
Contents
Environmental forecasting suites generate forecast products from a potentially large group of interdependent scientific models and associated data processing tasks. They are constrained by availability of external driving data: typically one or more tasks will wait on real time observations and/or model data from an external system, and these will drive other downstream tasks, and so on. The dependency diagram for a single forecast cycle point in such a system is a Directed Acyclic Graph as shown in Figure 1 (in our terminology, a forecast cycle point is comprised of all tasks with a common cycle point, which is the nominal analysis time or start time of the forecast models in the group). In real time operation processing will consist of a series of distinct forecast cycle points that are each initiated, after a gap, by arrival of the new cycle point’s external driving data.
From a job scheduling perspective task execution order in such a system must be carefully controlled in order to avoid dependency violations. Ideally, each task should be queued for execution at the instant its last prerequisite is satisfied; this is the best that can be done even if queued tasks are not able to execute immediately because of resource contention.
Cylc was developed for the EcoConnect Forecasting System at NIWA (National Institute of Water and Atmospheric Research, New Zealand). EcoConnect takes real time atmospheric and stream flow observations, and operational global weather forecasts from the Met Office (UK), and uses these to drive global sea state and regional data assimilating weather models, which in turn drive regional sea state, storm surge, and catchment river models, plus tide prediction, and a large number of associated data collection, quality control, preprocessing, post-processing, product generation, and archiving tasks.1 The global sea state forecast runs once daily. The regional weather forecast runs four times daily but it supplies surface winds and pressure to several downstream models that run only twice daily, and precipitation accumulations to catchment river models that run on an hourly cycle assimilating real time stream flow observations and using the most recently available regional weather forecast. EcoConnect runs on heterogeneous distributed hardware, including a massively parallel supercomputer and several Linux servers.
Most dependence between tasks applies within a single forecast cycle point. Figure 1 shows the dependency diagram for a single forecast cycle point of a simple example suite of three forecast models (a, b, and c) and three post processing or product generation tasks (d, e and f). A scheduler capable of handling this must manage, within a single forecast cycle point, multiple parallel streams of execution that branch when one task generates output for several downstream tasks, and merge when one task takes input from several upstream tasks.
|
|
Figure 2 shows the optimal job schedule for two consecutive cycle points of the example suite in real time operation, given execution times represented by the horizontal extent of the task bars. There is a time gap between cycle points as the suite waits on new external driving data. Each task in the example suite happens to trigger off upstream tasks finishing, rather than off any intermediate output or event; this is merely a simplification that makes for clearer diagrams.
|
|
|
|
Now the question arises, what happens if the external driving data for upcoming cycle points is available in advance, as it would be after a significant delay in operations, or when running a historical case study? While the forecast model a appears to depend only on the external data x at this stage of the discussion, in fact it would typically also depend on its own previous instance for the model background state used in initializing the new forecast. Thus, as alluded to in Figure 3, task a could in principle start as soon as its predecessor has finished. Figure 4 shows, however, that starting a whole new cycle point at this point is dangerous - it results in dependency violations in half of the tasks in the example suite. In fact the situation could be even worse than this - imagine that task b in the first cycle point is delayed for some reason after the second cycle point has been launched. Clearly we must consider handling inter-cycle-point dependence explicitly or else agree not to start the next cycle point early, as is illustrated in Figure 5.
Forecast models typically depend on their own most recent previous forecast for background state or restart files of some kind (this is called warm cycling) but there can also be inter-cycle-point dependence between different tasks. In an atmospheric forecast analysis suite, for instance, the weather model may generate background states for observation processing and data-assimilation tasks in the next cycle point as well as for the next forecast model run. In real time operation inter-cycle-point dependence can be ignored because it is automatically satisfied when one cycle point finishes before the next begins. If it is not ignored it drastically complicates the dependency graph by blurring the clean boundary between cycle points. Figure 6 illustrates the problem for our simple example suite assuming minimal inter-cycle-point dependence: the warm cycled models (a, b, and c) each depend on their own previous instances.
For this reason, and because we tend to see forecasting suites in terms of their real time characteristics, other metaschedulers have ignored inter-cycle-point dependence and are thus restricted to running entire cycle points in sequence at all times. This does not affect normal real time operation but it can be a serious impediment when advance availability of external driving data makes it possible, in principle, to run some tasks from upcoming cycle points before the current cycle point is finished - as was suggested at the end of the previous section. This can occur, for instance, after operational delays (late arrival of external data, system maintenance, etc.) and to an even greater extent in historical case studies and parallel test suites started behind a real time operation. It can be a serious problem for suites that have little downtime between forecast cycle points and therefore take many cycle points to catch up after a delay. Without taking account of inter-cycle-point dependence, the best that can be done, in general, is to reduce the gap between cycle points to zero as shown in Figure 5. A limited crude overlap of the single cycle point job schedule may be possible for specific task sets but the allowable overlap may change if new tasks are added, and it is still dangerous: it amounts to running different parts of a dependent system as if they were not dependent and as such it cannot be guaranteed that some unforeseen delay in one cycle point, after the next cycle point has begun, (e.g. due to resource contention or task failures) won’t result in dependency violations.
|
|
|
|
Figure 7 shows, in contrast to Figure 4, the optimal two cycle point job schedule obtained by respecting all inter-cycle-point dependence. This assumes no delays due to resource contention or otherwise - i.e. every task runs as soon as it is ready to run. The scheduler running this suite must be able to adapt dynamically to external conditions that impact on multi-cycle-point scheduling in the presence of inter-cycle-point dependence or else, again, risk bringing the system down with dependency violations.
|
|
|
|
To further illustrate the potential benefits of proper inter-cycle-point dependency handling, Figure 8 shows an operational delay of almost one whole cycle point in a suite with little downtime between cycle points. Above the time axis is the optimal schedule that is possible in principle when inter-cycle-point dependence is taken into account, and below it is the only safe schedule possible in general when it is ignored. In the former case, even the cycle point immediately after the delay is hardly affected, and subsequent cycle points are all on time, whilst in the latter case it takes five full cycle points to catch up to normal real time operation.
Similarly, Figure 9 shows example suite job schedules for an historical case study, or when catching up after a very long delay; i.e. when the external driving data are available many cycle points in advance. Task a, which as the most upstream forecast model is likely to be a resource intensive atmosphere or ocean model, has no upstream dependence on co-temporal tasks and can therefore run continuously, regardless of how much downstream processing is yet to be completed in its own, or any previous, forecast cycle point (actually, task a does depend on co-temporal task x which waits on the external driving data, but that returns immediately when the data is available in advance, so the result stands). The other forecast models can also cycle continuously or with a short gap between, and some post processing tasks, which have no previous-instance dependence, can run continuously or even overlap (e.g. e in this case). Thus, even for this very simple example suite, tasks from three or four different cycle points can in principle run simultaneously at any given time.
In fact, if our tasks are able to trigger off internal outputs of upstream tasks (message triggers) rather than waiting on full completion, then successive instances of the forecast models could overlap as well (because model restart outputs are generally completed early in the forecast) for an even more efficient job schedule.
|
|
Cylc manages a pool of proxy objects that represent the real tasks in a suite. Task proxies know how to run the real tasks that they represent, and they receive progress messages from the tasks as they run (usually reports of completed outputs). There is no global cycling mechanism to advance the suite; instead individual task proxies have their own private cycle point and spawn their own successors when the time is right. Task proxies are self-contained - they know their own prerequisites and outputs but are not aware of the wider suite. Inter-cycle-point dependence is not treated as special, and the task pool can be populated with tasks with many different cycle points. The task pool is illustrated in Figure 10. Whenever any task changes state due to completion of an output, every task checks to see if its own prerequisites have been satisfied. In effect, cylc gets a pool of tasks to self-organize by negotiating their own dependencies so that optimal scheduling, as described in the previous section, emerges naturally at run time.
The following packages are technically optional as you can construct and run cylc suites without dependency graphing, the gcylc GUI, or template processing but this is not recommended, and without Jinja2 you will not be able to run many of the example suites:
If you use a binary package manager to install graphviz you may also need a couple of devel packages for the pygraphviz build:
This user guide can be generated from the LATEXsource by running make in the top level cylc directory after download. The following TEXpackages are required (but note that the exact packages required may be somewhat OS or distribution-dependent):
And for HTML versions of the User Guide:
Finally, cylc makes heavy use of Python ordered dictionary data structures. Significant speedup in parsing large suites can be had by installing the fast C-coded ordereddict module by Anthon van der Neut:
This module is currently included with cylc under $CYLC_DIR/ext, and is built by the top level cylc Makefile. If you install the resulting library appropriately cylc will automatically use it in place of a slower Python implementation of the ordered dictionary structure.
Cylc should run “out of the box” on recent Linux distributions.
For distributed suites the Pyro versions installed on all suite or task hosts must be mutually compatible. Using identical Pyro versions guarantees compatibility but may not be strictly necessary because cylc uses Pyro rather minimally.
Beware of Linux distributions that come packaged with old Pyro versions. Pyro 3.9 and earlier is not compatible with the new-style Python classes used in cylc. It has been reported that Ubuntu 10.04 (Lucid Lynx), released in September 2009, suffers from this problem. Surprisingly, so does Ubuntu 11.10 (Oneiric Ocelot), released in October 2011 - and therefore, presumably, all earlier Ubuntu releases. Attempting to run a suite with Pyro 3.9 or earlier installed results in the following Python traceback:
It has been reported that cylc runs fine on OSX 10.6 SnowLeopard, but on OSX 10.7 Lion there is an issue with constructing proper FQDNs (Fully Qualified Domain Names) that requires a change to the DNS service. Here’s how to solve the problem:
Cylc has incorporated a custom-modified version the xdot graph viewer (http://code.google.com/p/jrfonseca/wiki/XDot, LGPL license).
First install Pyro, graphviz, Pygraphviz, Jinja2, TEX, and ImageMagick using the package manager on your system if possible; otherwise download the packages manually and follow their native installation documentation. On a modern Linux system, this is very easy. For example, to install cylc-5.1.0 on the Fedora 18 Linux distribution:
If you do not have root access on your intended cylc host machine and cannot get a sysadmin to do this at system level, see 4.4 for tips on installing everything to a local user account.
Now check that everything other than the LATEXpackages is installed properly:
If this command reports any errors then the packages concerned are not installed, not in the system Python search path, or (for a local install) not present in your $PYTHONPATH variable.
Cylc installs into a normal user account, as an unpacked release tarball or a git repository clone. See the INSTALL file in the source tree for instructions (also listed in G).
Site and user global config files define some important parameters that affect all suites, some of which may need to be customized for your site. See 6 for how to generate an initial site file and where to install it. All legal site and user global config items are defined in B.
The cylc test battery is primarily intended for developers to check that changes to the source code don’t break existing functionality. Note that some test failures can be expected to result from suites timing out, even if nothing is wrong, if you run too many tests in parallel. See cylc test-battery --help.
It is possible to install cylc and all of its software prerequisites under your own user account. Cylc itself is already designed to be installed into a normal user account, just follow the instructions above in 4.2. For the other packages, depending on the installation method used for each, it is just a matter of learning how to change the default install path prefix from, for example, /usr/local to $HOME/installed/usr/local and then ensuring that the resulting local package paths are set properly in your PYTHONPATH environment variable.
The graphviz build reportedly may fail on systems that do not have QT installed, hence the ./configure --with-qt=no option above. The graphviz lib and include locations are required when installing Pygraphviz.
Finally, check that everything (other than LATEXfor document processing) is installed:
If this command reports any errors then the packages concerned are not installed, not in the system Python search path, or (for a local install) not present in your $PYTHONPATH variable.
Upgrading is just a matter of unpacking the new cylc release. Successive cylc releases can be installed in parallel as suggested in the INSTALL file (G).
The problem with the word “cycle” is that it is often used to refer to a series of times (e.g. “an hourly cycle”) as well as a particular time (e.g. “current cycle” or “current cycle point”). Here, we use “cycle point” to refer to a particular (usually date-time) item. If a task has an hourly date-time cycle that starts at 2014-01-01T00 and ends at 2014-01-02T00, the first “cycle point” will be 2014-01-01T01, the second “cycle point” will be 2014-01-01T02, etc.
You may be accustomed to the idea that a forecasting suite has a “current cycle” (our nomenclature: “current cycle point”), which is typically the analysis time or nominal start time of the main forecast model(s) in the suite, and that the whole suite advances to the next forecast cycle point when all tasks in the current cycle point have finished (or even when a particular wall clock time is reached, in real time operation). As explained in the Introduction, this is not how cylc works.
Cylc suites advance by means of individual tasks with private cycle points (times) independently spawning successors at the next valid cycle point for the task, not by incrementing a suite-wide forecast cycle point. Each task will be submitted when its own prerequisites are satisfied, regardless of other tasks with other cycle points running, or not, at the time. It may still be convenient at times, however, to refer to the “current cycle point”, the “previous cycle point”, or the “next cycle point” and so forth, with reference to a particular task, or in the sense of all tasks that “belong to” a particular forecast cycle point. But keep in mind that the members of these groups may not be present simultaneously in the running suite - i.e. different tasks may pass through the “current cycle point” (etc.) at different times as the suite evolves, particularly in delayed (catch up) operation.
Cylc site and user global configuration files contain settings that affect all suites. Some of these, such as the range of network ports used by cylc, should be set at site level,
Others, such as the preferred text editor for suite definitions, can be overridden by users,
The cylc get-site-config command retrieves current global settings consisting of cylc defaults overridden by site settings, if any, overridden by user settings, if any. To generate an initial site or user global config file:
Settings that do not need to be changed should be deleted or commented out of user global config files so that they don’t override future changes to the site file.
Legal items, values, and system defaults are documented in the Site And User Config File Reference (B).
This section provides a hands-on tutorial introduction to basic cylc suite preparation and control. A number of features are not yet touched on by the tutorial examples, however, so please also read the rest of the User Guide.
Some global parameters affecting cylc’s behaviour are defined in a site global config file, and can be customized per user in user global config files. For example, to choose the text editor invoked by cylc on suite definitions:
Cylc has command line (CLI) and graphical (GUI) user interfaces. To get access to them you just need the cylc bin directory in your shell search path:
Note however that direct use of a specific installed version, like this, is not recommended. Instead a special wrapper script provided with cylc should be installed centrally for all users. This points to a top level cylc release directory and allows use of multiple cylc versions at once if necessary. See the cylc INSTALL file for details (G).
The command line interface is unified under a single top level cylc command that provides access to many sub-commands and their help documentation.
The cylc GUI covers the same functionality as the CLI, but it has more sophisticated suite monitoring capability. It can start and stop suites, or connect to suites that are already running; in either case, shutting down the GUI does not affect the suite itself.
Clicking on a suite in the summary GUI, shown in Figure 15, opens a gcylc instance for it.
Cylc suites are defined by extended-INI format suite.rc files (the main file format extension is section nesting). These reside in suite definition directories that may also contain a bin directory and any other suite-related files.
Suite registration associates a name with a suite definition directory, in a simple database. Cylc commands that parse suite definition files can take the file path or the suite name as input; commands that interact with running suites have to target the suite by name.
At registration time a random string of characters is written to a file called passphrase in the suite definition directory. At run time any contact from cylc client programs (running tasks, user commands, the cylc GUI) must use the same passphrase to authenticate with the running suite. This prevents unauthorized users from interfering with your suites (network communication between running processes is not subject to Unix user account permissions). Local tasks and user commands on the suite host automatically use the passphrase in the suite definition directory. For remote tasks and commands, however, the passphrase must be installed appropriately on the remote account - see 7.16 below.
Run the following command to import cylc’s example suites to a chosen directory location and register them for use under the examples name group:
(first check that $TMPDIR is defined in your environment, or else use a different location). List the newly registered tutorial suites using the cylc print command:
See cylc db print --help for other display options. The tree-form display shows how hierarchical suite names can be used to organize related suites nicely (suite names do not have to be related to their source directory paths, although they are in this case):
Suite definitions can be validated against the suite.rc file format specification to detect many types of error without running the suite.
Here’s the traditional Hello World program rendered as a cylc suite:
Cylc suites feature a clean separation of scheduling configuration, which determines when tasks are ready to run; and runtime configuration, which determines what to run (and where and how to run it) when a task is ready. In this example the [scheduling] section defines a single task called hello that triggers immediately when the suite starts up. When the task finishes the suite shuts down. That this is a dependency graph will be more obvious when more tasks are added. Under the [runtime] section the command scripting item defines a simple inlined implementation for hello: it sleeps for ten seconds, then prints Hello World!, and exits. This ends up in a job script generated by cylc to encapsulate the task (below) and, thanks to some defaults designed to allow quick prototyping of new suites, it is submitted to run as a background job on the suite host. In fact cylc even provides a default task implementation that makes the entire [runtime] section technically optional:
(the resulting dummy task just prints out some identifying information and exits).
The text editor invoked by cylc on suite definitions is determined by cylc site and user global config files, as shown above in 7.2. Check that you have renamed the tutorial examples suites as described just above and open the Hello World suite definition in your text editor:
Alternatively, start gcylc on the suite,
and choose Suite → Edit from the menu.
The editor will be invoked from within the suite definition directory for easy access to other suite files (in this case there are none). There are syntax highlighting control files for several text editors under /path/to/cylc/conf/; see in-file comments for installation instructions.
If you’re quick enough (this example only takes 10-15 seconds to run) the cylc scan command will detect the running suite:
Using the --no-detach option prevents the suite daemonizing, but only warnings and errors, if any, are printed to the terminal. The --debug option, however, causes (among other things) task job submission commands to to be printed stdout:
When a task is ready to run a job script is generated to run it. The command printed just above is how cylc executes background jobs on the suite host (this is the default), retrieves the job process ID, and directs the output to log files in standard locations. For the log file locations that can be seen above, in job/1/hello/01/ the first 1/ is the cycle point of the task hello (for a non-cycling task this is just the integer 1); and the final 01/ directory is its submission number (tasks can be made to retry on failure, manually or automatically, in which case the submission number increments).
The suite shuts down automatically once all tasks have succeeded.
The cylc GUI can start and stop suites, or (re)connect to suites that are already running:
Use the tool bar Play button, or the Control → Run menu item, to run the suite again. You may want to alter the suite definition slightly to make the task take longer to run. Try right-clicking on the hello task to view its output logs. The relative merits of the three suite views - dot, text, and graph - will be more apparent later when we have more tasks. Closing the GUI does not affect the suite itself.
Suites that are currently running can be detected with command line or GUI tools:
At run time, task instances are identified by name, which is determined entirely by the suite definition, and a cycle point which is usually a date-time or an integer:
Non-cycling tasks usually just have the cycle point 1, but this still has to be used to target the task instance with cylc commands.
Task job scripts are generated by cylc to wrap the task implementation specified in the suite definition (environment, command scripting, etc.) in error trapping code, messaging calls to report task progress back to the suite daemon, and so forth. Job scripts are written to the suite job log directory where they can be viewed alongside the job output logs. They can be accessed at run time by right-clicking on the task in the cylc GUI, or printed to the terminal:
This command can also print the suite log (and stdout and stderr for suites in daemon mode) and task stdout and stderr logs (see cylc log --help). A new job script can also be generated on the fly for inspection,
Take a look at the job script generated for hello.1 during the suite run above. The task command scripting should be clearly visible toward the bottom of the file.
The hello task in the first tutorial suite defaults to running as a background job on the suite host. To submit it to the Unix at scheduler instead, configure its job submission settings as in tut.oneoff.jobsub:
If you run the suite (first check that the at daemon atd is running on the suite host) you’ll see a slightly different job submission command printed to stdout:
Cylc supports a number of different job submission methods. Tasks submitted to external batch queuing systems like at, PBS, SLURM, or loadleveler, are displayed as submitted in the cylc GUI until they actually start executing.
If the --no-detach option is not used, suite stdout and stderr will be directed to the suite run directory along with the time-stamped suite log file, and task job scripts and job logs (task stdout and stderr). The default suite run directory location is $HOME/cylc-run:
The suite run database, suite environment file, suite state files, and task status files are used internally by cylc. Tasks execute in private work/ directories that are deleted automatically if empty when the task finishes. The suite share/ directory is made available to all tasks (by $CYLC_SUITE_SHARE_DIR) as a common share space. The task submission number increments from 1 if a task retries on failure; this is used a sub-directory of the log tree to avoid overwriting log files from earlier job submissions.
The top level run directory location can be changed in site and user config files if necessary, and the suite share and work locations can be configured separately because of the potentially larger disk space requirement.
Task job logs can be viewed by right-clicking on tasks in the gcylc GUI (so long as the task proxy is live in the suite), manually accessed from the log directory (of course), or printed to the terminal with the cylc log command:
The hello task in the first two tutorial suites defaults to running on the suite host. To make it run on a remote host instead change its runtime configuration as in tut.oneoff.remote:
For remote task hosting to work several requirements must be satisfied:
If your username is different on the task host the [[[remote]]] section also supports an owner=username item, or your $HOME/.ssh/config file can be configured for username translation.
If you configure a task host according to the requirements above and run the suite again in debug mode, you’ll see that the job submission command printed to suite stdout is now considerably more complicated. That’s because it has to create remote log directories, source login scripts on the remote host to ensure cylc is visible there, send the task job script over, and submit it to run there by the configured job submission method:
(Don’t be intimated by this - it really quite straightforward and would appear much simpler if the job log path was shorter!). Remote task job logs are saved to the suite run directory on the task host, not on the suite host, although they can be retrieved by right-clicking on the task in the GUI. Rose (section 14.1) provides a task event handler to pull logs back to the suite host.
To make a second task called goodbye trigger after hello finishes successfully, return to the original example, tut.oneoff.basic, and change the suite graph as in tut.oneoff.goodbye:
or to trigger it at the same time as hello,
and configure the new task’s behaviour under [runtime]:
Run tut.oneoff.goodbye and check the output from the new task:
Task names in the graph string can be qualified with a state indicator to trigger off task states other than success:
A common use of this is to automate recovery from known modes of failure:
i.e. if task goodbye fails, trigger another task that (presumably) really says goodbye.
Failure triggering generally requires use of suicide triggers as well, to remove the recovery task if it isn’t required (otherwise it would hang about indefinitely in the waiting state):
This means if goodbye fails, trigger really_goodbye; and otherwise, if goodbye succeeds, remove really_goodbye from the suite.
Try running tut.oneoff.suicide, which also configures the hello task’s runtime to make it fail, to see how this works.
The [runtime] section is actually a multiple inheritance hierarchy. Each subsection is a namespace that represents a task, or if it is inherited by other namespaces, a family. This allows common configuration to be factored out of related tasks very efficiently.
The [root] namespace provides defaults for all tasks in the suite. Here both tasks inherit command scripting from root, which they customize with different values of the environment variable $GREETING. Note that inheritance from root is implicit; from other parents an explicit inherit = PARENT is required, as shown below.
Task families defined by runtime inheritance can also be used as shorthand in graph trigger expressions. To see this, consider two “greeter” tasks that trigger off another task foo,
If we put the common greeting functionality of greeter_1 and greeter_2 into a special GREETERS family, the graph can be expressed more efficiently like this:
i.e. if foo succeeds, trigger all members of GREETERS at once. Here’s the full suite with runtime hierarchy shown:
Verbose validation shows the family member substitution done when the suite definition is parsed:
title = "Triggering a family of tasks"
[scheduling]
[[dependencies]]
graph = "foo => GREETERS"
[runtime]
[[root]]
pre-command scripting = "sleep 10"
[[foo]]
# empty (creates a dummy task)
[[GREETERS]]
command scripting = "echo $GREETING World!"
[[greeter_1]]
inherit = GREETERS
[[[environment]]]
GREETING = Hello
[[greeter_2]]
inherit = GREETERS
[[[environment]]]
GREETING = Goodbye
Experiment with the tut.oneoff.ftrigger1 suite to see how this works.
Tasks (or families) can also trigger off other families, but in this case we need to specify what the trigger means in terms of the upstream family members. Here’s how to trigger another task bar if all members of GREETERS succeed:
Verbose validation in this case reports:
Cylc ignores family member qualifiers like succeed-all on the right side of a trigger arrow, where they don’t make sense, to allow the two graph lines above to be combined in simple cases:
Any task triggering status qualified by -all or -any, for the members, can be used with a family trigger. For example, here’s how to trigger bar if all members of GREETERS finish (succeed or fail) and any of them them succeed:
(use of GREETERS:succeed-any by itself here would trigger bar as soon as any one member of GREETERS completed successfully). Verbose validation now begins to show how family triggers can simplify complex graphs, even for this tiny two-member family:
Experiment with tut.oneoff.ftrigger2 to see how this works.
You can style dependency graphs with an optional [visualization] section, as shown in tut.oneoff.ftrigger2:
To display the graph in an interactive viewer,
It should look like Figure 16 (with the GREETERS family node expanded on the right).
Graph styling can be applied to entire families at once, and custom “node groups” can also be defined for non-family groups.
The tasks in our examples so far have all had inlined implementation, in the suite definition, but real tasks often need to call external commands, scripts, or executables. To try this, let’s return to the basic Hello World suite and cut the implementation of the task hello out to a file hello.sh in the suite bin directory:
Make the task script executable, and change the hello task runtime section to invoke it:
If you run the suite now the new greeting from the external task script should appear in the hello task stdout log. This works because cylc automatically adds the suite bin directory to $PATH in the environment passed to tasks via their job scripts. To execute scripts (etc.) located elsewhere you can refer to the file by its full file path, or set $PATH appropriately yourself (this could be done via $HOME/.profile, which is sourced at the top of the task job script, or in the suite definition itself).
Note the use of set -e above to make the script abort on error. This allows the error trapping code in the task job script to automatically detect unforeseen errors.
So far we’ve considered non-cycling tasks, which finish without spawning a successor.
Cycling is based around iterating through date-time or integer sequences. A cycling task may run at each cycle point in a given sequence (cycle). For example, a sequence might be a set of date-times every 6 hours starting from a particular date-time. A cycling task may run for each date-time item (cycle point) in that sequence.
There may be multiple instances of this type of task running in parallel, if the opportunity arises and their dependencies allow it. Alternatively, a sequence can be defined with only one valid cycle point - in that case, a task belonging to that sequence may only run once.
Open the tut.cycling.one suite:
The difference between cycling and non-cycling suites is all in the [scheduling] section, so we will leave the [runtime] section alone for now (this will result in cycling dummy tasks). Note that the graph is now defined under a new section heading that makes each task under it have a succession of cycle points ending in 00 or 12 hours, between specified initial and final cycle points (or indefinitely if no final cycle point is given), as shown in Figure 17.
title = "Two cycling tasks, no inter-cycle dependence"
[cylc]
UTC mode = True
[scheduling]
initial cycle point = 20130808T00
final cycle point = 20130812T00
[[dependencies]]
[[[T00,T12]]] # 00 and 12 hours UTC every day
graph = "foo => bar"
[visualization]
initial cycle point = 20130808T00
final cycle point = 20130809T00
[[node attributes]]
foo = "color=red"
bar = "color=blue"
If you run this suite instances of foo will spawn in parallel out to the runahead limit, and each bar will trigger off the corresponding instance of foo at the same cycle point. The runahead limit, which defaults to a few cycles but is configurable, prevents uncontrolled spawning of cycling tasks in suites that are not constrained by clock triggers in real time operation.
Experiment with tut.cycling.one to see how cycling tasks work.
ISO 8601 Date-Time Syntax The suite above is a very simple example of a cycling date-time workflow. More generally, cylc comprehensively supports the ISO 8601 standard for date-time instants, intervals, and sequences. Cycling graph sections can be specified using full ISO 8601 recurrence expressions, but these may be simplified by assuming context information from the suite - namely initial and final cycle points. One form of the recurrence syntax looks like Rn/start-date-time/period (Rn means repeat n times). In the example above, if the initial cycle point is always at 00 or 12 hours then [[[T00,T12]]] could be written as [[[PT12H]]], which is short for [[[R/initial-cycle-point/PT12H/]]] - i.e. repeat indefinitely every 12 hours starting at the initial cycle point.
The tut.cycling.two suite adds inter-cycle-point dependence to the previous example:
For any given cycle point in the sequence defined by the cycling graph section heading, bar triggers off foo as before, but now foo triggers off its own previous instance foo[-PT12H]. Date-time offsets in inter-cycle-point triggers are expressed as ISO 8601 intervals (12 hours in this case). Figure 18 shows how this connects the cycling graph sections together.
Experiment with this suite to see how inter-cycle-point triggers work. Note that the first instance of foo, at suite start-up, will trigger immediately in spite of its inter-cycle-point trigger, because cylc ignores dependence on points earlier than the initial cycle point. However, the presence of an inter-cycle-point trigger usually implies something special has to happen at start-up. If a model depends on its own previous instance for restart files, for example, then some special process has to generate the initial set of restart files when there is no previous cycle point to do it. The following section shows one way to handle this in cylc suites.
Sometimes we want to be able to run a task at the initial cycle point, but refrain from running it in subsequent cycles. We can do this by writing an extra set of dependencies that are only valid at a single date-time cycle point. If we choose this to be the initial cycle point, these will only apply at the very start of the suite.
The cylc syntax for writing this single date-time cycle point occurrence is R1, which stands for R1/no-specified-date-time/no-specified-period. This is an adaptation of part of the ISO 8601 date-time standard’s recurrence syntax (Rn/date-time/period) with some special context information supplied by cylc for the no-specified-⋆ data.
The 1 in the R1 means repeat once. As we’ve specified no date-time, Cylc will use the initial cycle point date-time by default, which is what we want. We’ve also missed out specifying the period - this is set by cylc to a zero amount of time in this case (as it never repeats, this is not significant).
For example, in tut.cycling.three:
This is shown in Figure 19.
Note that the time zone has been set to +1300 in this case, instead of UTC (Z) as before. If no time zone or UTC mode was set, the local time zone of your machine will be used in the cycle points.
At the initial cycle point, foo will depend on foo[-PT12H] and also on prep:
Thereafter, it will just look like e.g.:
However, in our initial cycle point example, the dependence on foo.20130807T1200+13 will be ignored, because that task’s cycle point is earlier than the suite’s initial cycle point and so it cannot run. This means that the initial cycle point dependencies for foo actually look like:
Cylc can do also do integer cycling for repeating workflows that are not date-time based.
Open the tut.cycling.integer suite, which is plotted in Figure 20.
[scheduling]
cycling mode = integer
initial cycle point = 1
final cycle point = 3
[[dependencies]]
[[[R1]]] # = R1/1/?
graph = start => foo
[[[P1]]] # = R/1/P1
graph = foo[-P1] => foo => bar
[[[R2/P1]]] # = R2/P1/3
graph = bar => stop
[visualization]
[[node attributes]]
start = "style=filled", "fillcolor=skyblue"
foo = "style=filled", "fillcolor=slategray"
bar = "style=filled", "fillcolor=seagreen3"
stop = "style=filled", "fillcolor=orangered"
The integer cycling notation is intended to look similar to the ISO 8601 date-time notation, but it is simpler for obvious reasons. The example suite illustrates two recurrence forms, Rn/start-point/period and Rn/period/stop-point, simplified somewhat using suite context information (namely the initial and final cycle points). The first form is used to run one special task called start at start-up, and for the main cycling body of the suite; and the second form to run another special task called stop in the final two cycles. The P character denotes period (interval) just like in the date-time notation. R/1/P2 would generate the sequence of points 1,3,5,....
Cylc has built in support for the Jinja2 template processor, which allows us to embed code in suite definitions to generate the final result seen by cylc.
The tut.oneoff.jinja2 suite illustrates two common uses of Jinja2: changing suite content or structure based on the value of a logical switch; and iteratively generating dependencies and runtime configuration for groups of related tasks:
To view the result of Jinja2 processing with the Jinja2 flag MULTI set to False:
#!jinja2
{% set MULTI = True %}
{% set N_GOODBYES = 3 %}
title = "A Jinja2 Hello World! suite"
[scheduling]
[[dependencies]]
{% if MULTI %}
graph = "hello => BYE"
{% else %}
graph = "hello"
{% endif %}
[runtime]
[[hello]]
command scripting = "sleep 10; echo Hello World!"
{% if MULTI %}
[[BYE]]
command scripting = "sleep 10; echo Goodbye World!"
{% for I in range(0,N_GOODBYES) %}
[[ goodbye_{{I}} ]]
inherit = BYE
{% endfor %}
{% endif %}
And with MULTI set to True:
Tasks can be configured to retry a number of times if they fail. An environment variable $CYLC_TASK_TRY_NUBMER increments from 1 on each successive try, and is passed to the task to allow different behaviour on the retry:
When a task with configured retries fails, its cylc task proxy goes into the retrying state until the next retry delay is up, then it resubmits. It only enters the failed state on a final definitive failure.
title = "A task with automatic retry on failure"
[scheduling]
[[dependencies]]
graph = "hello"
[runtime]
[[hello]]
retry delays = 2⋆0.1 # retry twice after 0.1 minute delays
command scripting = """
sleep 10
if [[ $CYLC_TASK_TRY_NUMBER < 3 ]]; then
echo "Hello ... aborting!"
exit 1
else
echo "Hello World!"
fi"""
Experiment with tut.oneoff.retry to see how this works.
If you have read access to another user’s account (even on another host) it is possible to use cylc monitor to look at their suite’s progress without full shell access to their account. To do this, you will need to copy their suite passphrase to
(use of the host and owner names is optional here - see 12.6.1) and also retrieve the port number of the running suite from,
Once you have this information, you can run
to view the progress of their suite.
Other suite-connecting commands work in the same way too; see 12.10.
The cylc suite search tool reports pattern matches in the suite definition by line number, suite section, and file, even if the suite uses nested include-files, and by file and line number for matches in suite bin scripts:
Almost every feature of cylc can be tested quickly and easily with a simple dummy suite. You can write your own, or start from one of the example suites in /path/to/cylc/examples (see use of cylc import-examples above) - they all run “out the box” and can be copied and modified at will.
Cylc commands target suites via names registered in a suite name database located at $HOME/.cylc/REGDB/. Suite names are hierarchical like directory paths, allowing nested tree-like grouping, but use the ‘.’ character as a delimiter. This :
Suite titles held in the name database are parsed from the suite definition at the time of initial suite registration. If you change the title later use cylc db refresh to update the database.
Name groups are entirely virtual, they do not need to be explicitly created before use, and they automatically disappear if all tasks are removed from them. From the listing above, for example, to move the suite nwp.oper.region2 into the nwp.test group:
And to move nwp.test.region2 into a new group nwp.para:
Currently you cannot explicitly indicate a group name on the command line by appending a dot character. Rather, in database operations such as copy, reregister, or unregister, the identity of the source item (group or suite) is inferred from the content of the database; and if the source item is a group, so must the target be a group (or it will be, in the case of an item that will be created by the operation). This means that you cannot copy a single suite into a group that does not exist yet unless you specify the entire target suite name.
cylc db register --help shows a number of other examples.
On the command line, the ‘database’ (or ‘db’) command category contains commands to implement the aforementioned operations.
Groups of suites (at any level in the name hierarchy) can be deleted, copied, imported, and exported; as well as individual suites. To do this, just use suite group names as source and/or target for operations, as appropriate. For instance, if a group foo.bar contains the suites foo.bar.baz and foo.bar.qux, you can copy a single suite like this:
(resulting in a new suite boo); or the group like this:
(resulting in new suites boo.baz and boo.qux); or the group like this:
(resulting in new suites boo.bar.baz and boo.bar.qux). When suites are copied, the suite definition directories are copied into a directory tree, under the target directory, that reflects the suite name hierarchy. cylc copy --help has some explicit examples.
The same functionality is also available by right-clicking on suites or groups in the gcylc “Open Registered Suite” dialog.
Any client process that connects to a running suite (this includes task messaging and user-invoked interrogation and control commands) must authenticate with a secure passphrase that has been loaded by the suite. A random passphrase is generated automatically in the suite definition directory at registration time if one does not already exist there. For the default Pyro-based connection method the passphrase file must be distributed to other accounts that host running tasks or from which you need monitoring or control access to the running suite.
Alternatively, cylc can be configured to,
Neither of these methods require the suite passphrase to be installed on the task host. For ssh re-invocation ssh keys must be installed for the task-to-suite direction in addition to the suite-to-task setup already required for job submission. The automatic polling mechanism can be used as a last resort for hosts that do not allow routing back to the suite host for pyro or ssh. It can also be used as regular health check on submitted tasks under the other communications methods.
See 12 for more detail on cylc client/server communications, and how to use it.
Cylc suites are defined in structured, validated, suite.rc files that concisely specify the properties of, and the relationships between, the various tasks managed by the suite. This section of the User Guide deals with the format and content of the suite.rc file, including task definition. Task implementation - what’s required of the real commands, scripts, or programs that do the processing that the tasks represent - is covered in 10; and task job submission - how tasks are submitted to run - is in 11.
A cylc suite definition directory contains:
A typical example:
Suite.rc files are an extended-INI format with section nesting.
Embedded template processor expressions may also be used in the file, to programatically generate the final suite definition seen by cylc. Currently the Jinja2 template processor is supported (http://jinja.pocoo.org/docs); see 9.6 for examples. In the future cylc may provide a plug-in interface to allow use of other template engines too.
The following defines legal suite.rc syntax:
Suites that embed Jinja2 code (see 9.6) must process to raw suite.rc syntax.
Cylc has native support for suite.rc include-files, which may help to organize large suites. Inclusion boundaries are completely arbitrary - you can think of include-files as chunks of the suite.rc file simply cut-and-pasted into another file. Include-files may be included multiple times in the same file, and even nested. Include-file paths can be specified portably relative to the suite definition directory, e.g.:
Editing Temporarily Inlined Suites Cylc’s native file inclusion mechanism supports optional inlined editing:
The suite will be split back into its constituent include-files when you exit the edit session. While editing, the inlined file becomes the official suite definition so that changes take effect whenever you save the file. See cylc prep edit --help for more information.
Include-Files via Jinja2 Jinja2 (9.6) also has template inclusion functionality.
Cylc comes with syntax files for a number of text editors:
Refer to comments at the top of each file to see how to use them.
Cylc suite.rc files consist of a suite title and description followed by configuration items grouped under several top level section headings:
Cylc suite.rc files are automatically validated against a specification that defines all legal entries, values, options, and defaults. This detects formatting errors, typographic errors, illegal items and illegal values prior to run time. Some values are complex strings that require further parsing by cylc to determine their correctness (this is also done during validation). All legal entries are documented in the Suite.rc Reference (A).
The validator reports the line numbers of detected errors. Here’s an example showing a section heading with a missing right bracket:
If the suite.rc file uses include-files cylc view will show an inlined copy of the suite with correct line numbers (you can also edit suites in a temporarily inlined state with cylc edit --inline).
Validation does not check the validity of chosen job submission methods.
The [scheduling] section of a suite.rc file defines the relationships between tasks in a suite - the information that allows cylc to determine when tasks are ready to run. The most important component of this is the suite dependency graph. Cylc graph notation makes clear textual graph representations that are very concise because sections of the graph that repeat at different hours of the day, say, only have to be defined once. Here’s an example with dependencies that vary depending on the particular cycle point:
Figure 21 shows the complete suite.rc listing alongside the suite graph. This is a complete, valid, runnable suite (it will use default task runtime properties such as command scripting).
|
Multiline graph strings may contain:
Suite dependency graphs can be broken down into pairs in which the left side (which may be a single task or family, or several that are conditionally related) defines a trigger for the task or family on the right. For instance the “word graph” C triggers off B which triggers off A can be deconstructed into pairs C triggers off B and B triggers off A. In this section we use only the default trigger type, which is to trigger off the upstream task succeeding; see 9.3.4 for other available triggers.
In the case of cycling tasks, the triggers defined by a graph string are valid for cycle points matching the list of hours specified for the graph section. For example this graph,
implies that B triggers off A for cycle points in which the hour matches 00 or 12.
To define inter-cycle-point dependencies, attach an offset indicator to the left side of a pair:
This means B[time] triggers off A[time-PT12H] (12 hours before) for cycle points with hours matching 00 or 12. time is implicit because this keeps graphs clean and concise, given that the majority of tasks will typically depend only on others with the same cycle point. Cycle point offsets can only appear on the left of a pair, because a pairs define triggers for the right task at cycle point time. However, A => B[-PT6H], which is illegal, can be reformulated as a future trigger A[+PT6H] => B (see 9.3.4.10). It is also possible to combine multiple offsets within a cycle point offset e.g.
This means that B[Time] triggers off A[time-P1D-PT12H] (1 day and 12 hours before).
Triggers can be chained together. This graph:
is equivalent to this:
Each trigger in the graph must be unique but the same task can appear in multiple pairs or chains. Separately defined triggers for the same task have an AND relationship. So this:
is equivalent to this:
In summary, the branching tree structure of a dependency graph can be partitioned into lines (in the suite.rc graph string) of pairs or chains, in any way you like, with liberal use of internal white space and comments to make the graph structure as clear as possible.
Handling Long Graph Lines Long chains of dependencies can be split into pairs:
If you have very long task names, or long conditional trigger expressions (below) then you can use the suite.rc line continuation marker:
Note that a line continuation marker must be the final character on the line; it cannot be followed by trailing spaces or a comment.
A suite definition can contain multiple graph strings that are combined to generate the final graph.
One-off (Non-Cycling) Figure 22 shows a small suite of one-off non-cycling tasks; these all share a single cycle point (1) and don’t spawn successors (once they’re all finished the suite just exits). The integer 1 attached to each graph node is just an arbitrary label here.
Cycling Graphs For cycling tasks the graph section heading defines a sequence of cycle points for which the subsequent graph section is valid. Figure 23 shows a small suite of cycling tasks.
Advanced Cycling The hour-based section heading examples shown above (T00, T06, etc) are just one small part of cylc’s cycling syntax.
T06 stands for ”Repeat every 1 day starting at 06:00 after the initial cycle point”. Cylc allows you to start (or end) at any particular time, repeat at whatever frequency you like, and even optionally limit the number of repetitions.
For example, all these are valid in cylc:
Here is a quick summary of the syntax rules:
Date-time cycling information is made up of a starting date-time, an interval, and an optional limit.
The time is assumed to be in the local time zone unless you set [cylc]cycle point time zone or [cylc]UTC mode. The calendar is assumed to be the proleptic Gregorian calendar unless you set [scheduling]cycling mode.
The syntax for representations is based on the ISO 8601 date-time standard. This includes the representation of date-time, interval. What we define for cylc’s cycling syntax is our own optionally-heavily-condensed form of ISO 8601 recurrence syntax. The most common full form is: R[limit?]/[date-time]/[interval]. However, we allow omitting information that can be guessed from the context (rules below). This means that it can be written as:
For example, T00 is an example of [date-time], with an inferred 1 day period and no limit.
Where some or all date-time information is omitted, it is inferred to be relative to the initial date-time cycle point. For example, T00 by itself would mean the next occurrence of midnight that follows, or is, the initial cycle point. Entering +PT6H would mean 6 hours after the initial cycle point. Entering -P1D would mean 1 day before the initial cycle point. Entering no information for the date-time implies the initial cycle point date-time itself.
Where the interval is omitted and some (but not all) date-time information is omitted, it is inferred to be a single unit above the largest given specific date-time unit. For example, the largest given specific unit in T00 is hours, so the inferred interval is 1 day (daily), P1D.
Where the limit is omitted, unlimited cycling is assumed. This will be bounded by the final cycle point’s date-time if given.
Another supported form of ISO 8601 recurrence is: R[limit?]/[interval]/[date-time]. This form uses the date-time as the end of the cycling sequence rather than the start. For example, R3/P5D/20140430T06 means:
This kind of form can be used for specifying special behaviour near the end of the suite, at the final cycle point’s date-time. We can also represent this in cylc with a collapsed form:
So, for example, you can write:
How Multiple Graph Strings Combine For a cycling graph with multiple validity sections for different hours of the day, the different sections add to generate the complete graph. Different graph sections can overlap (i.e. the same hours may appear in multiple section headings) and the same tasks may appear in multiple sections, but individual dependencies should be unique across the entire graph. For example, the following graph defines a duplicate prerequisite for task C:
This does not affect scheduling, but for the sake of clarity and brevity the graph should be written like this:
Advanced Starting Up Dependencies that are only valid at the initial cycle point can be written using the R1 notation (e.g. as in 7.23.2. For example:
In the example above, R1 implies R1/20130808T00, so prep only runs once at that cycle point (the initial cycle point). At that cycle point, foo will have a dependence on prep - but not at subsequent cycle points.
However, it is possible to have a suite that has multiple effective initial cycles - for example, one starting at T00 and another starting at T12. What if they need to share an initial task?
Let’s suppose that we add the following section to the suite example above:
We’ll also say that there should be a starting dependence between prep and our new task baz - but we still want to have a single prep task, at a single cycle.
We can write this using a special case of the task[-interval] syntax - if the interval is null, this implies the task at the initial cycle point.
For example, we can write our suite like 24.
|
This neatly expresses what we want - a task running at the initial cycle point that has one-off dependencies with other task sets at different cycles.
Integer Cycling In addition to non-repeating and date-time cycling workflows, cylc can do integer cycling for repeating workflows that are not date-time based.
To construct an integer cycling suite, set [scheduling]cycling mode = integer, and specify integer values for the initial and (optional) final cycle points. The notation for intervals, offsets, and recurrences (sequences) is similar to the date-time cycling notation, except for the simple integer values.
The full integer recurrence expressions supported are:
But, as for date-time cycling, sequence start and end points can be omitted where suite initial and final cycle points can be assumed. Some examples:
The tutorial illustrates integer cycling in 7.23.3, and $CYLC_DIR/examples/satellite/ is a self-contained example of a realistic use for integer cycling. It simulates the processing of incoming satellite data: each new dataset arrives after a random (as far as the suite is concerned) interval, and is labeled by an arbitrary (as far as the suite is concerned) ID in the filename. A task called get_data at the top of the repeating workflow waits on the next dataset and, when it finds one, moves it to a cycle-point-specific shared workspace for processing by the downstream tasks. When get_data.1 finishes, get_data.2 triggers and begins waiting for the next dataset at the same time as the downstream tasks in cycle point 1 are processing the first one, and so on. In this way multiple datasets can be processed at once if they happen to come in quickly. A single shutdown task runs at the end of the final cycle to collate results. The suite graph is shown in Figure 25.
Trigger type, indicated by :type after the upstream task (or family) name, determines what kind of event results in the downstream task (or family) triggering.
Success Triggers The default, with no trigger type specified, is to trigger off the upstream task succeeding:
For consistency and completeness, however, the success trigger can be explicit:
Failure Triggers To trigger off the upstream task reporting failure:
Suicide triggers can be used to remove task B here if A does not fail, see 9.3.4.8.
Start Triggers To trigger off the upstream task starting to execute:
This can be used to trigger tasks that monitor other tasks once they (the target tasks) start executing. Consider a long-running forecast model, for instance, that generates a sequence of output files as it runs. A postprocessing task could be launched with a start trigger on the model (model:start => post) to process the model output as it becomes available. Note, however, that there are several alternative ways of handling this scenario: both tasks could be triggered at the same time (foo => model & post), but depending on external queue delays this could result in the monitoring task starting to execute first; or a different postprocessing task could be triggered off a message output for each data file (model:out1 => post1 etc.; see 9.3.4.5), but this may not be practical if the number of output files is large or if it is difficult to add cylc messaging calls to the model.
Finish Triggers To trigger off the upstream task succeeding or failing, i.e. finishing one way or the other:
Message Triggers Message triggers allow triggering off custom messages emitted by a task as it runs. Message outputs must be registered for the task in the suite definition, and matching messages sent back to the suite daemon by the task at the appropriate time. Note that polling does not yet detect custom message output completion. Custom output messages should normally contain the cycle point in order to distinguish between the outputs of different instances of the same task. The task implementation can use $CYLC_TASK_CYCLE_POINT for this, or cylc cycle-point --offset=P2M for an offset value. The matching message string registered in the suite definition, however, does not get interpreted by the shell, so a cycle point placeholder is used instead: [] for the current cycle point, or for an offset [-P2M].
The $CYLC_DIR/examples/message-triggers/ suite is a self-contained example that illustrates message triggering. Note that the graph trigger notation uses a label that selects the message registered under the task [runtime] section.
title = "test suite for cylc-6 message triggers"
[scheduling]
initial cycle point = 20140801T00
final cycle point = 20141201T00
[[dependencies]]
[[[P2M]]]
graph = """
# bar triggers off message 'x' emitted by foo:
foo:x => bar
# baz triggers off message 'y' emitted by the previous instance of foo:
foo[-P2M]:y => baz
"""
[runtime]
[[foo]]
command scripting = """
echo HELLO
sleep 2
TARGET_POINT=$CYLC_TASK_CYCLE_POINT
cylc message "file 1 for $TARGET_POINT done"
sleep 2
TARGET_POINT=$(cylc cycle-point --offset P2M)
cylc message "file 2 for $TARGET_POINT done"
sleep 2"""
[[[outputs]]]
x = "file 1 for [] done"
y = "file 2 for [P2M] done"
Job Submission Triggers It is also possible to trigger off a task submitting, or failing to submit:
A possible use case for submit-fail triggers: if a task goes into the submit-failed state, possibly after several job submission retries, another task that inherits the same runtime but sets a different job submission method and/or host could be triggered to, in effect, run the same job on a different platform.
Conditional Triggers AND operators (&) can appear on both sides of an arrow. They provide a concise alternative to defining multiple triggers separately:
OR operators (|) which result in true conditional triggers, can only appear on the left,2
Forecasting suites typically have simple conditional triggering requirements, but any valid conditional expression can be used, as shown in Figure 26 (conditional triggers are plotted with open arrow heads).
Suicide Triggers Suicide triggers take tasks out of the suite. This can be used for automated failure recovery. The suite.rc listing and accompanying graph in Figure 27 show how to define a chain of failure recovery tasks that trigger if they’re needed but otherwise remove themselves from the suite (you can run the AutoRecover.async example suite to see how this works). The dashed graph edges ending in solid dots indicate suicide triggers, and the open arrowheads indicate conditional triggers as usual.
|
Note that multiple suicide triggers combine in the same way as other triggers, so this:
is equivalent to this:
i.e. both foo and bar must succeed for baz to be taken out of the suite. If you really want a task to be taken out if any one of several events occurs then be careful to write it that way:
A word of warning on the meaning of “bare suicide triggers”. Consider the following suite:
Task bar has a suicide trigger but no normal prerequisites (a suicide trigger is not a task triggering prerequisite, it is a task removal prerequisite) so this is entirely equivalent to:
In other words both tasks will trigger immediately, at the same time, and then bar will be removed if foo succeeds.
If an active task proxy (currently in the submitted or running states) is removed from the suite by a suicide trigger, a warning will be logged.
Family Triggers Families defined by the namespace inheritance hierarchy ( 9.4) can be used in the graph trigger whole groups of tasks at the same time (e.g. forecast model ensembles and groups of tasks for processing different observation types at the same time) and for triggering downstream tasks off families as a whole. Higher level families, i.e. families of families, can also be used, and are reduced to the lowest level member tasks. Note that tasks can also trigger off individual family members if necessary.
To trigger an entire task family at once:
This is equivalent to:
To trigger other tasks off families we have to specify whether to triggering off all members starting, succeeding, failing, or finishing, or off any members (doing the same). Legal family triggers are thus:
Here’s how to trigger downstream processing after if one or more family members succeed, but only after all members have finished (succeeded or failed):
Inter-Cycle-Point Triggers Typically most tasks in a suite will trigger off others in the same cycle point, but some may depend on others with other cycle points. This notably applies to warm-cycled forecast models, which depend on their own previous instances (see below); but other kinds of inter-cycle-point dependence are possible too.3 Here’s how to express this kind of relationship in cylc:
inter-cycle-point and trigger type (or message trigger) notation can be combined:
At suite start-up inter-cycle-point triggers refer to a previous cycle point that does not exist. This does not cause the dependent task to wait indefinitely, however, because cylc ignores triggers that reach back beyond the initial cycle point. That said, the presence of an inter-cycle-point trigger does normally imply that something special has to happen at start-up. If a model depends on its own previous instance for restart files, for instance, then an initial set of restart files has to be generated somehow or the first model task will presumably fail with missing input files. There are several ways to handle this in cylc using different kinds of one-off (non-cycling) tasks that run at suite start-up. They are illustrated in the Tutorial (7.23.1); to summarize here briefly:
R1, or R1/date-time tasks are the recommended way to specify unusual start up conditions. They allow you to specify a clean distinction between the dependencies of initial cycles and the dependencies of the subsequent cycles.
Initial tasks can be used for real model cold-start processes, whereby a warm-cycled model at any given cycle point can in principle have its inputs satisfied by a previous instance of itself, or by an initial task with (nominally) the same cycle point.
In effect, the R1 task masquerades as the previous-cycle-point trigger of its associated cycling task. At suite start-up initial tasks will trigger the first cycling tasks, and thereafter the inter-cycle-point trigger will take effect.
If a task has a dependency on another task in a different cycle point, the dependency can be written using the [offset] syntax such as [-PT12H] in foo[-PT12H] => foo. This means that foo at the current cycle point depends on a previous instance of foo at 12 hours before the current cycle point. Unlike the cycling section headings (e.g. [[[T00,T12]]]), dependencies assume that relative times are relative to the current cycle point, not the initial cycle point.
However, it can be useful to have specific dependencies on tasks at or near the initial cycle point. You can switch the context of the offset to be the initial cycle point by using the caret symbol: ^.
For example, you can write foo[^] to mean foo at the initial cycle point, and foo[^+PT6H] to mean foo 6 hours after the initial cycle point. Usually, this kind of dependency will only apply in a limited number of cycle points near the start of the suite, so you may want to write it in R1-based cycling sections. Here’s the example inter-cycle R1 suite from above again.
You can see there is a dependence on the initial R1 task prep for foo at the first T00 cycle point, and at the first T12 cycle point. Thereafter, foo just depends on its previous (12 hours ago) instance.
This is quite different to the non-preferred cold-start tasks way of working.
Special Cold-Start Tasks Special cold-start tasks are non-spawning tasks intended for use in general cycling sections, so they must be declared as special tasks. Dependence on them is retained throughout the suite run so they must be used in conditional triggers to avoid stalling the suite, like this:
Here, in the first cycle point model will trigger off the cold-start task cold_model. In subsequent cycle points model retains its dependence on cold_model, but the conditional expression allows it to trigger off the previous instance of itself instead. The reason for doing this is that it allows cold_model to be inserted manually later in the suite run to cold-start model again after problems that require skipping one or more cycles. If you do not need this mid-run cold start capability then it is simpler to start the suite with a normal task in an initial R1 recurrence, like this:
Special Sequential Tasks If a cycling task does not generate files required by its own successor, then successive instances can run in parallel if the opportunity arises. However, if such a task would interfere with its own siblings for internal reasons (e.g. use of a hardwired non cycle dependent temporary file or similar) then it can be forced to run sequentially. This can be done with explicit inter-cycle-point triggers in the graph:
or by declaring the task to be sequential:
The sequential declaration also results in each instance of foo triggering off its own predecessor, exactly as in the explicit version. The only difference is that implicit triggers will not appear in graph visualizations. The implicit version can also be considerably simpler when the task appears in multiple graph sections or in a non-uniform cycling sequence: this suite,
is equivalent to this one:
Future Triggers Cylc also supports inter-cycle-point triggering off tasks in the future (with respect to cycle point date-time, not real time!):
In contrast to normal inter-cycle-point triggers, future triggers present a problem at the suite stop time rather than at start-up - in the final cycle point B wants to to trigger off A at a future cycle point that does not exist. To avoid this problem cylc prevents tasks from spawning successors that depend on tasks in a non-existent future cycle point.
Warm-cycled forecast models generate restart files, e.g. model background fields, to initialize the next forecast. This kind of dependence requires an inter-cycle-point trigger:
Or if you don’t need an explicit cold-start task, this will do:
If your model is configured to write out additional restart files to allow one or more cycle points to be skipped in an emergency do not represent these potential dependencies in the suite graph as they should not be used under normal circumstances. For example, the following graph would result in task A erroneously triggering off A[T-24] as a matter of course, instead of off A[T-6], because A[T-24] will always be finished first:
If you need to skip one or more cycle points in a suite like this, manually remove the tasks that cannot run (or use cylc purge to remove their downstream dependents too) and manually trigger the first post-gap task.
The [runtime] section of a suite definition configures what to execute (and where and how to execute it) when each task is ready to run, in a multiple inheritance hierarchy of namespaces culminating in individual tasks. This allows all common configuration detail to be factored out and defined in one place.
Any namespace can configure any or all of the items defined in the Suite.rc Reference (A).
Namespaces that do not explicitly inherit from others automatically inherit from the root namespace (below).
Nested namespaces define task families that can be used in the graph as convenient shorthand for triggering all member tasks at once, or for triggering other tasks off all members at once - see 9.3.4.9. Nested namespaces can be progressively expanded and collapsed in the dependency graph viewer, and in the gcylc graph and text views. Only the first parent of each namespace (as for single-inheritance) is used for suite visualization purposes.
Namespace names may contain letters, digits, underscores, and hyphens.
Note that task names need not be hardwired into task implementations because task and suite identity can be extracted portably from the task execution environment supplied by the suite daemon (9.4.7) - then to rename a task you can just change its name in the suite definition.
The root namespace, at the base of the inheritance hierarchy, provides default configuration for all tasks in the suite. Most root items are unset by default, but some have default values sufficient to allow test suites to be defined by dependency graph alone. The command scripting item, for example, defaults to code that prints a message then sleeps for between 1 and 15 seconds and exits. Default values are documented with each item in A. You can override the defaults or provide your own defaults by explicitly configuring the root namespace.
If a namespace section heading is a comma-separated list of names then the subsequent configuration applies to each list member. Particular tasks can be singled out at run time using the $CYLC_TASK_NAME variable.
As an example, consider a suite containing an ensemble of closely related tasks that each invokes the same script but with a unique argument that identifies the calling task name:
For large ensembles Jinja2 template processing can be used to automatically generate the member names and associated dependencies (see 9.6).
The following listing of the inherit.single.one example suite illustrates basic runtime inheritance with single parents.
# SUITE.RC
title = "User Guide [runtime] example."
[cylc]
required run mode = simulation # (no task implementations)
[scheduling]
initial cycle point = 20110101T06
final cycle point = 20110102T00
[[dependencies]]
[[[T00]]]
graph = """foo => OBS
OBS:succeed-all => bar"""
[runtime]
[[root]] # base namespace for all tasks (defines suite-wide defaults)
[[[job submission]]]
method = at_now
[[[environment]]]
COLOR = red
[[OBS]] # family (inherited by land, ship); implicitly inherits root
command scripting = run-${CYLC_TASK_NAME}.sh
[[[environment]]]
RUNNING_DIR = $HOME/running/$CYLC_TASK_NAME
[[land]] # a task (a leaf on the inheritance tree) in the OBS family
inherit = OBS
description = land obs processing
[[ship]] # a task (a leaf on the inheritance tree) in the OBS family
inherit = OBS
description = ship obs processing
[[[job submission]]]
method = loadleveler
[[[environment]]]
RUNNING_DIR = $HOME/running/ship # override OBS environment
OUTPUT_DIR = $HOME/output/ship # add to OBS environment
[[foo]]
# (just inherits from root)
# The task [[bar]] is implicitly defined by its presence in the
# graph; it is also a dummy task that just inherits from root.
If a namespace inherits from multiple parents the linear order of precedence (which namespace overrides which) is determined by the so-called C3 algorithm used to find the linear method resolution order for class hierarchies in Python and several other object oriented programming languages. The result of this should be fairly obvious for typical use of multiple inheritance in cylc suites, but for detailed documentation of how the algorithm works refer to the official Python documentation here: http://www.python.org/download/releases/2.3/mro/.
The inherit.multi.one example suite, listed here, makes use of multiple inheritance:
cylc get-suite-config provides an easy way to check the result of inheritance in a suite. You can extract specific items, e.g.:
title = "multiple inheritance example"
description = """To see how multiple inheritance works:
% cylc list -tb[m] SUITE # list namespaces
% cylc graph -n SUITE # graph namespaces
% cylc graph SUITE # dependencies, collapse on first-parent namespaces
% cylc get-config --sparse --item [runtime]ops_s1 SUITE
% cylc get-config --sparse --item [runtime]var_p2 foo"""
[scheduling]
[[dependencies]]
graph = "OPS:finish-all => VAR"
[runtime]
[[root]]
[[OPS]]
command scripting = echo "RUN: run-ops.sh"
[[VAR]]
command scripting = echo "RUN: run-var.sh"
[[SERIAL]]
[[[directives]]]
job_type = serial
[[PARALLEL]]
[[[directives]]]
job_type = parallel
[[ops_s1, ops_s2]]
inherit = OPS, SERIAL
[[ops_p1, ops_p2]]
inherit = OPS, PARALLEL
[[var_s1, var_s2]]
inherit = VAR, SERIAL
[[var_p1, var_p2]]
inherit = VAR, PARALLEL
[visualization]
# NOTE ON VISUALIZATION AND MULTIPLE INHERITANCE: overlapping
# family groups can have overlapping attributes, so long as
# non-conflictling attributes are used to style each group. Below,
# for example, OPS tasks are filled green and SERIAL tasks are
# outlined blue, so that ops_s1 and ops_s2 are green with a blue
# outline. But if the SERIAL tasks are explicitly styled as "not
# filled" (by setting "style=") this will override the fill setting
# in the (previously defined and therefore lower precedence) OPS
# group, making ops_s1 and ops_s2 unfilled with a blue outline.
# Alternatively you can just create a manual node group for ops_s1
# and ops_s2 and style them separately.
[[node groups]]
#(see comment above:)
#serial_ops = ops_s1, ops_s2
[[node attributes]]
OPS = "style=filled", "fillcolor=green"
SERIAL = "color=blue" #(see comment above:), "style="
#(see comment above:)
#serial_ops = "color=blue", "style=filled", "fillcolor=green"
or use the --sparse option to print entire namespaces without obscuring the result with the dense runtime structure obtained from the root namespace:
Suite Visualization And Multiple Inheritance The first parent inherited by a namespace is also used as the collapsible family group when visualizing the suite. If this is not what you want, you can demote the first parent for visualization purposes, without affecting the order of inheritance of runtime properties:
The linear precedence order of ancestors is computed for each namespace using the C3 algorithm. Then any runtime items that are explicitly configured in the suite definition are “inherited” up the linearized hierarchy for each task, starting at the root namespace: if a particular item is defined at multiple levels in the hierarchy, the level nearest the final task namespace takes precedence. Finally, root namespace defaults are applied for every item that has not been configured in the inheritance process (this is more efficient than carrying the full dense namespace structure through from root from the beginning).
The task execution environment contains suite and task identity variables provided by the suite daemon, and user-defined environment variables. The environment is explicitly exported (by the task job script) prior to executing task command scripting (see 11).
Suite and task identity are exported first, so that user-defined variables can refer to them. Order of definition is preserved throughout so that variable assignment expressions can safely refer to previously defined variables.
Additionally, access to cylc itself is configured prior to the user-defined environment, so that variable assignment expressions can make use of cylc utility commands:
User Environment Variables A task’s user-defined environment results from its inherited [[[environment]]] sections:
This results in a task foo with SHAPE=circle, COLOR=blue, and TEXTURE=rough in its environment.
Overriding Environment Variables When you override inherited namespace items the original parent item definition is replaced by the new definition. This applies to all items including those in the environment sub-sections which, strictly speaking, are not “environment variables” until they are written, post inheritance processing, to the task job script that executes the associated task. Consequently, if you override an environment variable you cannot also access the original parent value:
The compressed variant of this, COLOR = dark-$COLOR, is also in error for the same reason. To achieve the desired result you must use a different name for the parent variable:
Suite And Task Identity Variables The task identity variables provided to tasks by the suite daemon include:
And the suite identity variables are:
Some of these variables are also used by cylc task messaging commands in order to target the right task proxy object in the right suite.
Suite Share And Task Work Directories A suite share directory is created automatically for use as a file exchange area for tasks on same task host. It can be accessed via $CYLC_SUITE_SHARE_DIR and its location can be set in the cylc site and user global config files.
A task work directory is also created automatically for each task, and can be accessed via the $CYLC_TASK_WORK_DIR variable. Task command scripting is executed from within the work directory (i.e. it is the task’s current working directory). For non-detaching tasks the work directory is automatically removed again if it is empty when the task finishes. The main work directory location is set in the cylc site and user global config files, but the lowest-level sub-directory, which name defaults to the task ID to give each task a unique workspace, can be overridden under [runtime] in suite definitions. This enables groups of tasks that read and write files from their current working directories to be given common work directories as file share spaces.
Other Cylc-Defined Environment Variables Initial and final cycle points, if supplied via the suite.rc file or the command line, are passed to task execution environments as:
Tasks can use these to determine whether or not they are running in the first or final cycles. Note however that R1 graph sections are now the preferred way to get different behaviour at suite start-up or shutdown.
Environment Variable Evaluation Variables in the task execution environment are not evaluated in the shell in which the suite is running prior to submitting the task. They are written in unevaluated form to the job script that is submitted by cylc to run the task (11.1) and are therefore evaluated when the task begins executing under the task owner account on the task host. Thus $HOME, for instance, evaluates at run time to the home directory of task owner on the task host.
Tasks can use $CYLC_SUITE_DEF_PATH to access suite files on the task host, and the suite bin directory is automatically added $PATH. If a remote suite definition directory is not specified the local (suite host) path will be assumed with the local home directory, if present, swapped for literal $HOME for evaluation on the task host.
If a task declares an owner other than the suite owner and/or a host other than the suite host, cylc will use passwordless ssh to execute the task on the owner@host account by the configured job submission method,
For this to work,
To learn how to give remote tasks access to cylc, see 12.7.
Tasks running on the suite host under another user account are treated as remote tasks.
Remote hosting, like all namespace settings, can be declared globally in the root namespace, or per family, or for individual tasks.
Dynamic Host Selection Instead of hardwiring host names into the suite definition you can specify a shell command that prints a hostname, or an environment variable that holds a hostname, as the value of the host config item. See A.4.1.19.1.
Remote Task Log Directories Task stdout and stderr streams are written to log files in a suite-specific sub-directory of the suite run directory, as explained in 11.3. For remote tasks the same directory is used, but on the task host. Remote task log directories, like local ones, are created on the fly, if necessary, during job submission.
The visualization section of a suite definition is used to configure suite graphing, principally graph node (task) and edge (dependency arrow) style attributes. Tasks can be grouped for the purpose of applying common style attributes. See A for details.
Nested families from the runtime inheritance hierarchy can be expanded and collapsed in suite graphs and the gcylc graph view. All families are displayed in the collapsed state at first, unless [visualization]collapsed families is used to single out specific families for initial collapsing.
In the gcylc graph view, nodes outside of the main graph (such as the members of collapsed families) are plotted as rectangular nodes to the right if they are doing anything interesting (submitted, running, failed).
Figure 28 illustrates successive expansion of nested task families in the namespaces example suite.
|
Cylc has built in support for the Jinja2 template processor in suite definitions. Jinja2 variables, mathematical expressions, loop control structures, conditional logic, etc., are automatically processed to generate the final suite definition seen by cylc.
The need for Jinja2 processing must be declared with a hash-bang comment as the first line of the suite.rc file:
Potential uses for this include automatic generation of repeated groups of similar tasks and dependencies, and inclusion or exclusion of entire suite sections according to the value of a single flag. Consider a large complicated operational suite and several related parallel test suites with slightly different task content and structure (the parallel suites, for instance, might take certain large input files from the operation or the archive rather than downloading them again) - these can now be maintained as a single master suite definition that reconfigures itself according to the value of a flag variable indicating the intended use.
Template processing is the first thing done on parsing a suite definition so Jinja2 expressions can appear anywhere in the file (inside strings and namespace headings, for example).
Jinja2 is well documented at http://jinja.pocoo.org/docs, so here we just provide an example suite that uses it. The meaning of the embedded Jinja2 code should be reasonably self-evident to anyone familiar with standard programming techniques.
The jinja2.ensemble example, graphed in Figure 29, shows an ensemble of similar tasks generated using Jinja2:
Here is the generated suite definition, after Jinja2 processing:
And finally, the jinja2.cities example uses variables, includes or excludes special cleanup tasks according to the value of a logical flag, and it automatically generates all dependencies and family relationships for a group of tasks that is repeated for each city in the suite. To add a new city and associated tasks and dependencies simply add the city name to list at the top of the file. The suite is graphed, with the New York City task family expanded, in Figure 30.
#!Jinja2
title = "Jinja2 city suite example."
description = """
Illustrates use of variables and math expressions, and programmatic
generation of groups of related dependencies and runtime properties."""
{% set HOST = "SuperComputer" %}
{% set CITIES = 'NewYork', 'Philadelphia', 'Newark', 'Houston', 'SantaFe', 'Chicago' %}
{% set CITYJOBS = 'one', 'two', 'three', 'four' %}
{% set LIMIT_MINS = 20 %}
{% set CLEANUP = True %}
[scheduling]
initial cycle point = 2020-01-01 # Equivalent to 20200101.
[[ dependencies ]]
{% if CLEANUP %}
[[[T23]]]
graph = "clean"
{% endif %}
[[[T00,T12]]]
graph = """
setup => get_lbc & get_ic # foo
{% for CITY in CITIES %} {# comment #}
get_lbc => {{ CITY }}_one
get_ic => {{ CITY }}_two
{{ CITY }}_one & {{ CITY }}_two => {{ CITY }}_three & {{ CITY }}_four
{% if CLEANUP %}
{{ CITY }}_three & {{ CITY }}_four => cleanup
{% endif %}
{% endfor %}
"""
[runtime]
[[on_{{ HOST }} ]]
[[[remote]]]
host = {{ HOST }}
# (remote cylc directory is set in site/user config for this host)
[[[directives]]]
wall_clock_limit = "00:{{ LIMIT_MINS|int() + 2 }}:00,00:{{ LIMIT_MINS }}:00"
{% for CITY in CITIES %}
[[ {{ CITY }} ]]
inherit = on_{{ HOST }}
{% for JOB in CITYJOBS %}
[[ {{ CITY }}_{{ JOB }} ]]
inherit = {{ CITY }}
{% endfor %}
{% endfor %}
[visualization]
initial cycle point = 2011080812
final cycle point = 2011080823
[[node groups]]
cleaning = clean, cleanup
[[node attributes]]
cleaning = 'style=filled', 'fillcolor=yellow'
NewYork = 'style=filled', 'fillcolor=lightblue'
This functionality is not provided by Jinja2 by default, but cylc automatically imports the user environment to the template in a dictionary structure called environ. A usage example:
This example is emphasizes that the environment is read on the suite host at the time the suite definition is parsed - it is not, for instance, read at task run time on the task host.
Jinja2 variable values can be modified by “filters”, using pipe notation. For example, the built-in trim filter strips leading and trailing white space from a string:
(See official Jinja2 documentation for available built-in filters.)
Cylc also supports custom Jinja2 filters. A custom filter is a single Python function in a source file with the same name as the function (plus “.py” extension) and stored in one of the following locations:
In the filter function argument list, the first argument is the variable value to be “filtered”, and subsequent arguments can be whatever is needed. Currently there is one custom filter called “pad” in the central cylc Jinja2 filter directory, for padding string values to some constant length with a fill character - useful for generating task names and related values in ensemble suites:
Associative arrays (dicts in Python) can be very useful. Here’s an example, from
$CYLC_DIR/examples/jinja2/dict:
Here’s the result:
The values of Jinja2 variables can be passed in from the cylc command line rather than hardwired in the suite
definition. Here’s an example, from
$CYLC_DIR/examples/jinja2/defaults:
Here’s the result:
Note also that cylc view --set FIRST_TASKbob –jinja2 SUITE= will show the suite with the Jinja2 variables as set.
Warning: suites started with template variables set on the command line do not currently restart with the same settings - you have to set them again on the cylc restart command line.
It is sometimes convenient to omit certain tasks from the suite at runtime without actually deleting their definitions from the suite.
Defining [runtime] properties for tasks that do not appear in the suite graph results in verbose-mode validation warnings that the tasks are disabled. They cannot be used because the suite graph is what defines their dependencies and valid cycle points. Nevertheless, it is legal to leave these orphaned runtime sections in the suite definition because it allows you to temporarily remove tasks from the suite by simply commenting them out of the graph.
To omit a task from the suite at runtime but still leave it fully defined and available for use (by insertion or cylc submit) use one or both of [scheduling][[special task]] lists, include at start-up or exclude at start-up (documented in A.3.7.6 and A.3.7.5). Then the graph still defines the validity of the tasks and their dependencies, but they are not actually loaded into the suite at start-up. Other tasks that depend on the omitted ones, if any, will have to wait on their insertion at a later time or otherwise be triggered manually.
Finally, with Jinja2 (9.6) you can radically alter suite structure by including or excluding tasks from the [scheduling] and [runtime] sections according to the value of a single logical flag defined at the top of the suite.
A naked dummy task appears in the suite graph but has no explicit runtime configuration section. Such tasks automatically inherit the default “dummy task” configuration from the root namespace. This is very useful because it allows functional suites to be mocked up quickly for test and demonstration purposes by simply defining the graph. It is somewhat dangerous, however, because there is no way to distinguish an intentional naked dummy task from one generated by typographic error: misspelling a task name in the graph results in a new naked dummy task replacing the intended task in the affected trigger expression; and misspelling a task name in a runtime section heading results in the intended task becoming a dummy task itself (by divorcing it from its intended runtime config section).
To avoid this problem any dummy task used in a real suite should not be naked - i.e. it should have an explicit entry in under the runtime section of the suite definition, even if the section is empty. This results in exactly the same dummy task behaviour, via implicit inheritance from root, but it allows use of cylc validate --strict to catch errors in task names by failing the suite if any naked dummy tasks are detected.
Existing scripts or executables can be used as cylc tasks without any modification, unless:
Simple tasks can be entirely implemented within the suite.rc file - task command scripting can be a multi-line string.
Tasks should abort with non-zero exit status if a fatal error occurs (this is just good coding practice anyway). This allows cylc’s task job scripts to automatically trap errors and send a cylc task failed message back to the suite. The shell set -e option can be used in lieu of explicit error checks for every command:
General (non-output) messages can also be sent to report progress, warnings, and so on, e.g.:
Explanatory messages can be sent before aborting on error:
Or equivalently, with different syntax:
But not this:
If critical errors are not reported in this way task failures will still be detected and logged by the suite daemon, but you may have to examine task logs to determine what the problem was.
If a task spawns another job internally and then detaches and exits without seeing the spawned process through, you must arrange for the detached process to send its own completion messages, because the cylc-generated job script cannot know when it is finished.
First check that you can’t “reconnect” the detaching process. If it is a background shell process, for instance, just run it in the foreground instead. For loadleveler jobs the -s option prevents llsubmit from returning until the job has completed. For Sun Grid Engine, qsub -sync yes has the same effect. For how to override the job submission command template see 11.4.
If the detaching process cannot be reconnected, disable cylc’s automatic completion messaging:
The cylc messaging commands are called like this:
They read environment variables that identify the calling task and the target suite, so the task execution environment must be propagated to the detached process.
One way to handle this is to write a task wrapper that modifies a copy of the detaching native job scripts, on the fly, to insert completion messaging in the appropriate places. An advantage of this method is that you don’t need to permanently modify the model or its associated native scripting for cylc. Another is that you can configure the native job setup for a single test case (running it without cylc) and then have your custom wrapper modify the standalone test case on the fly with suite, task, and cycle-specific parameters as required.
To make this easier, for tasks that declare manual completion messaging cylc makes non user-defined environment scripting available in a variable $CYLC_SUITE_ENVIRONMENT, the value of which can be inserted at the appropriate point in the task scripts (just prior to calling the cylc messaging commands as above).4
Another reason to avoid detaching tasks if possible is that they cannot be polled or killed because there is no way for cylc to determine the job ID of the detached process. Attempted polling of a detaching task will just result in cylc logging a warning message.
The detaching example suite contains a script model.sh that runs a pseudo model as follows:
this is in turn executed by a script run-model.sh that detaches immediately after job submission (i.e. it exits before the model executable actually runs):
Note that your at scheduler daemon must be up if you want to test this suite.
#!/bin/bash
set -e
echo "run-model.sh: submitting model.sh to 'at now'"
SCRIPT=model.sh # location of the model job to submit
OUT=$1; ERR=$2 # stdout and stderr log paths
# submit the job and detach
MY_TMPDIR=${CYLC_TMPDIR:-${TMPDIR:-/tmp}}
RES=$MY_TMPDIR/atnow$$.txt
( at now <<EOF
$SCRIPT 1> $OUT 2> $ERR
EOF
) > $RES 2>&1
if grep 'No atd running' $RES; then
echo 'ERROR: atd is not running!'
exit 1
fi
# model.sh should now be running at the behest of the 'at' scheduler.
echo "run-model.sh: done"
Here’s a cylc suite to run this unruly model:
The suite invokes the task by means of the custom wrapper model-wrapper.sh which modifies, on the fly, a temporary copy of the model’s native job scripts as described above:
title = "Cylc User Guide Custom Task Wrapper Example"
description = """This suite runs a single task that internally submits a
'model executable' before detaching and exiting immediately - so we have
to handle task completion messaging manually - see the Cylc User Guide."""
[scheduling]
initial cycle point = 20110101T06
final cycle point = 20110102T00
[[special tasks]]
sequential = model
[[dependencies]]
[[[T00,T06,T12,T18]]]
graph = "model"
[runtime]
[[model]]
manual completion = True
command scripting = model-wrapper.sh # invoke the task via a custom wrapper
[[[environment]]]
# location of native job scripts to modify for this suite:
NATIVESCRIPTS = $CYLC_SUITE_DEF_PATH/native
# output path prefix for detached model stdout and stderr:
PREFIX = $CYLC_TASK_LOG_ROOT
FOO = "$HOME bar $PREFIX"
If you run this suite, or submit the model task alone with cylc submit, you’ll find that the usual job submission log files for task stdout and stderr end before the task is finished. To see the “model” output and the final task completion message (success or failure), examine the log files generated by the job submitted internally to the at scheduler (their location is determined by the $PREFIX variable in the suite.rc file).
#/bin/bash
set -eu
# A custom wrapper for the 'model' task from the detaching example suite.
# See the Cylc User Guide for more information.
MY_TMPDIR=${CYLC_TMPDIR:-${TMPDIR:-/tmp}}
# Get a temporary copy of the native job scripts:
TDIR=$MY_TMPDIR/detach$$
mkdir -p $TDIR
cp $NATIVESCRIPTS/⋆ $TDIR
# Insert task-specific execution environment in $TDIR/model.sh:
SRCH='echo "model.sh: executing pseudo-executable"'
perl -pi -e "s@^${SRCH}@${CYLC_SUITE_ENVIRONMENT}\n${SRCH}@" $TDIR/model.sh
# Task completion message scripting. Use single quotes here - we don't
# want the $? variable to evaluate in this shell!
MSG='
if [[ $? != 0 ]]; then
cylc task message -p CRITICAL "ERROR: model executable failed"
exit 1
else
cylc task succeeded
exit 0
fi'
# Insert error detection and cylc messaging in $TDIR/model.sh:
SRCH='echo "model.sh: done"'
perl -pi -e "s@^${SRCH}@${MSG}\n${SRCH}@" $TDIR/model.sh
# Point to the temporary copy of model.sh, in run-model.sh:
SRCH='SCRIPT=model.sh'
perl -pi -e "s@^${SRCH}@SCRIPT=$TDIR/model.sh@" $TDIR/run-model.sh
# Execute the (now modified) native process:
$TDIR/run-model.sh ${PREFIX}-detached.out ${PREFIX}-detached.err
echo "model-wrapper.sh: see modified job scripts under ${TDIR}!"
# EOF
It should not be difficult to adapt this example to real tasks with detaching internal job submission. You will probably also need to replace other parameters, such as model input and output filenames, with suite- and cycle-appropriate values, but exactly the same technique can be used: identify which job script needs to be modified and use text processing tools (such as the single line perl search-and-replace expressions above) to do the job.
For the requirements a command, script, or program, must fulfill in order to function as a cylc task, see 10. This section explains how tasks are submitted by the suite daemon when they are ready to run, and how to define new task job submission methods.
When a task is ready to run cylc generates a temporary task job script to configure the task’s execution environment and call its command scripting. The job script is the embodiment of all suite.rc runtime settings for the task. It is submitted to run by the job submission method configured for the task. Different tasks can have different job submission methods. Like other runtime properties, you can set a suite default job submission method and override it for specific tasks or families:
As shown in the Tutorial (7.11) job scripts are saved to the suite run directory; the commands used to submit them are printed to stdout by cylc in debug mode; and they can be printed with the cylc log command or new ones generated and printed with the cylc jobscript command. Take a look at one to see exactly how cylc wraps and runs your tasks.
Cylc supports a number of commonly used job submission methods. See 11.7 for how to add new job submission methods.
Runs tasks directly in a background shell.
Submits tasks to the rudimentary Unix at scheduler. The atd daemon must be running.
Submits tasks to loadleveler by the llsubmit command. Loadleveler directives can be provided in the suite.rc file:
These are written to the top of the task job script like this:
If restart=yes is specified as a directive for loadleveler, the job will automatically trap SIGUSR1, which loadleveler may use to preempt the job. On trapping SIGUSR1, the job will inform the suite that it has been vacated by loadleveler. This will put it back to the submitted state, until it starts running again.
Submits tasks to PBS (or Torque) by the qsub command. PBS directives can be provided in the suite.rc file:
These are written to the top of the task job script like this:
Submits tasks to Sun/Oracle Grid Engine by the qsub command. SGE directives can be provided in the suite.rc file:
These are written to the top of the task job script like this:
Submits tasks to Simple Linux Utility for Resource Management by the sbatch command. SLURM directives can be provided in the suite.rc file (note that since not all SLURM commands have a short form, cylc requires the long form directives):
These are written to the top of the task job script like this:
For job submission methods that use job file directives (PBS, Loadlevler, etc.) default directives are provided to set the job name and stdout and stderr file paths.
As shown in the example above, multiple entries for the same PBS or SGE directive option must be comma-separated on the same line, in the suite.rc file. Otherwise, repeating the option on another line will override the previous entry, not add to it. To get a naked option flag such as -cwd in SGE you must give a quoted blank space as the directive value in the suite.rc file.
When a task is ready to run cylc generates a filename root to be used for the task job script and log files. The filename containing the task name, cycle point, and a submit number that increments if the same task is re-triggered multiple times:
How the stdout and stderr streams are directed into these files depends on the job submission method. The background method just uses appropriate output redirection on the command line, as shown above. The loadleveler method writes appropriate directives to the job script that is submitted to loadleveler.
Cylc obviously has no control over the stdout and stderr output from tasks that do their own internal output management (e.g. tasks that submit internal jobs and direct the associated output to other files). For less internally complex tasks, however, the files referred to here will be complete task job logs.
To change the form of the actual command used to submit a job you do not need to define a new job submission method; just override the command template in the relevant job submission sections of your suite.rc file:
As explained in A the template’s first %s will be substituted by the job file path and, where applicable a second and third %s will be substituted by the paths to the job stdout and stderr files.
For supported job submission methods, one-way polling can be used to determine actual job status: the suite daemon executes a process on the task host, by passwordless ssh, to interrogate the batch queueing system there, and to read a status file that is automatically generated by the task job script as it runs.
Polling may be required to update the suite state correctly after unusual events such as a machine being rebooted with tasks running on it, or network problems that prevent task messages from getting back to the suite host.
Tasks can be polled on demand by right-clicking on them in gcylc or using the cylc poll command.
Tasks are polled automatically, once, if they timeout during job submission or execution (see A.4.1.20 for how to configure timeouts).
Any tasks recorded in the submitted or running states at suite restart are automatically polled to determine what happened to them while the suite was down.
Regular polling can also be configured as a health check on tasks submitted to hosts that are known to be flaky, or as the sole method of determining task status on hosts that do not allow task messages to be routed back to the suite host.
To use polling instead of task-to-suite messaging set task communication method = poll in cylc site and user global config (see B.8.1.3). The default polling intervals can be overridden for all suites there too (see B.1.8 and B.1.7), or in specific suite definitions (in which case polling will be done regardless of the task communication method configured for the host; see A.4.1.10 and A.4.1.11).
Note that regular polling is not as efficient as task messaging in updating task status, and it should be used sparingly in large suites.
For supported job submission methods, the suite daemon can execute a process on the task host, by passwordless ssh, to kill a submitted or running job according to its job submission method.
Tasks can be killed on demand by right-clicking on them in gcylc or using the cylc kill command.
Defining a new handler for a new job submission method requires a little Python programming. You can derive (in the sense of object oriented programming inheritance) new classes from one of the existing ones, or simply write a new one using the existing ones as examples. Full reference can be found in the header cylc.batch_sys_manager module.
The following user-defined job submission class, called qsub, overrides the built-in pbs class to change the directive prefix from #PBS to #QSUB:
To check that this works correctly save the new source file to qsub.py in one of the allowed locations (see just below), use it in a suite definition:
and generate a job script to see the resulting directives:
You new job submission class code should be saved to a file with the same name as the class (plus “.py” extension). It can reside in any of the following locations, depending on how generally useful the new method is and whether or not you have write-access to the cylc source tree:
Note that the form of the import statement at the top of the new user-defined Python module differs depending on whether or not the file is installed in the cylc source tree (see the comment at the top of the example file above).
This chapter currently features a diverse collection of topics related to running suites. Please also see the Tutorial (7) and command documentation (D), and experiment with plenty of examples.
There are three ways to start a suite running: cold start and warm start, which start from scratch; and restart, which loads a prior suite state. If a suite does not contain any special cold-start tasks there is no difference between cold and warm start, except that a warm start should start from a point beyond the original initial cycle point of the suite. Special cold-start tasks are only needed if you want to be able to cold-start individual tasks again in the middle of a suite run - see 9.3.4.11.
Once a suite is up and running it is typically a restart that is needed most often (but see also cylc reload). Be aware that cold and warm starts wipe out any prior suite state, which prevents returning to a restart if you decide that’s what you really intended.
A cold start is the primary way to start a suite run from scratch:
The initial cycle point may be specified on the command line or in the suite.rc file. The scheduler starts by loading the first instance of each task at the suite initial cycle point, or at the next valid point for the task, including any special cold-start tasks (see the note on cold-start tasks at the beginning of this section).
A restart starts a suite run from the state recorded at the end of a previous run. This allows restarting a suite that was shut down or killed, without rerunning tasks that were already completed, or which were already submitted or running when the suite went down.
For a restart, the scheduler starts by loading each task in its recorded state. The most recent state is loaded by default, but earlier state files can be specified on the command line. Any tasks recorded as ‘submitted’ or ‘running’ will be polled automatically to determine what happened to them while the suite was down.
A warm start runs a suite from scratch like a cold start, but from a given cycle point that is later than the suite’s initial cycle point. It can be considered an inferior alternative to a restart because it may result in some tasks rerunning. A warm start may be required if a restart is not possible because the suite state files were accidentally deleted (for instance). The warm start cycle point must be given on the command line:
In this case the scheduler starts by loading a first instance of each task at the warm start cycle point, or at the next valid point for the task, and any special cold-start tasks are loaded in the ‘succeeded’ state (this is merely a device to satisfy any initial dependence on the cold-start tasks, which are assumed not to be needed for a warm start; see the note on cold-start tasks at the beginning of this section).
Cylc has three ways of tracking the progress of tasks, configured per task host in the site and user global config files (6). All three methods can be used on different task hosts within the same suite if necessary.
The Pyro communication method is the default because it is the most direct and efficient; the ssh method inserts an extra step in the process (command re-invocation on the suite host); and task polling is the least efficient because results are checked at predetermined intervals, not when task events actually occur.
Be careful to avoid spamming task hosts with polling commands. Each poll opens (and then closes) a new ssh connection.
Polling intervals are configurable under [runtime] because they should may depend on the expected execution time. For instance, a task that typically takes an hour to run might be polled every 10 minutes initially, and then every minute toward the end of its run. Interval values are used in turn until the last value, which is used repeatedly until finished:
A list of intervals with optional multipliers can be used for both submission and execution polling, although a single value is probably sufficient for submission polling. If these items are not configured default values from site and user global config will be used for the polling task communication method; polling is not done by default under the other task communications methods (but it can still be used if you like).
Polling is also done automatically once on job submission and execution timeouts, to see if the timed-out task has failed or not; and on suite restarts, to see what happened to any tasks that were orphaned when the suite went down.
If Pyro and ssh ports are blocked but you don’t want to use polling from the suite host,
Here are the default site and user global config items relevant to task state tracking (see these with cylc get-site-config):
User-invoked commands that connect to running suites can also choose between direct communication across network sockets (Pyro) and re-invocation of commands on the suite host using passwordless ssh (there is a --use-ssh command option for this purpose).
The gcylc GUI requires direct Pyro connections to its target suite. If that is not possible, run gcylc on the suite host.
All Pyro connections to a running suite (task messaging and user-invoked commands) must authenticate with an arbitrary single line of text in a file called passphrase, which will be found and used automatically if installed properly - see below. A secure MD5 checksum, not the raw passphrase, is passed across the network. A random passphrase is generated in the suite definition directory when a suite is registered, but you can create your own if you wish.
For ssh task messaging and user command re-invocation, on the other hand, the suite passphrase is only required on the suite host account but ssh keys must be installed for passwordless connections instead.
Suite passphrases currently have to be installed manually to all task host accounts that use the Pyro communication method (see above); and also to accounts used to run commands that interact directly with the suite via Pyro.
Legal passphrase locations, in order of preference, are:
Remote tasks know the location of the remote suite definition directory (if one exists) through their execution environment. Local (suite host) user command invocations can find the suite definition directory in the suite name database. Remote user command invocations, however, cannot interrogate the database on the command host because the suite will not be registered there (cylc cannot assume that the command host shares a common filesystem with the suite host). Consequently remote command host accounts must have the suite passphrase installed in one of the secondary locations under $HOME/.cylc/.
Running tasks need access to cylc via $PATH, principally for the task messaging commands. To allow this, the first thing a task job script does is set $CYLC_VERSION to the cylc version number of the running suite. If you need to run several suites at once under different incompatible versions of cylc, check that your site is using the cylc version wrapper (see INSTALL and admin/cylc-wrapper in a cylc installation) then set $CYLC_VERSION to the desired version. In the case of developers wishing to run their own copy of cylc rather than a centrally installed one, set $CYLC_HOME to point to your cylc copy.
Access to the cylc executable for different hosts can be configured using the site and user global configuration files. If the environment for running the cylc executable is only set up correctly in a login shell for a given host, you can set the [hosts]$\rightarrow$ HOST $\rightarrow$ cylc executable setting for the relevant host to True. (This is the default behaviour.) If the environment is already correct without the login shell, but the cylc executable is not in $PATH, then the [hosts]$\rightarrow$ HOST $\rightarrow$ cylc executable setting can be used to specify the path to the cylc executable.
A restarted suite (see cylc restart --help) is initialized from a previous recorded suite state dump so that it can carry on from wherever it got to before being shut down or killed.
Tasks that were recorded in the submitted or running states are now automatically polled on restart, to see if they are still submitted (e.g. waiting in a PBS batch queue or similar), still running, or if they finished (succeeded or failed) while the suite was down.
Tasks recorded in the failed state at shutdown are not automatically resubmitted on restarting the suite, in case the underlying problem has not been addressed yet.
As a suite runs its task proxies may pass through the following states:
Note that greyed-out “base graph nodes” in the gcylc graph view do not represent task states; they are displayed to fill out the graph structure where corresponding task proxies do not currently exist in the live task pool.
For manual task state reset purposes ready is a pseudo-state that means waiting with all prerequisites satisfied.
Connecting to a running suite requires knowing the network port it is listening on, and the suite passphrase to authenticate with once a connection is made to the port.
Suites write their port number to $HOME/.cylc/ports/<SUITE> at start-up, and suite-connecting commands read this file to get the number.5 An exception to this is the messaging commands called by tasks. Running tasks know the port number from the execution environment provided by the suite (via the task job script).
So, to connect to a suite running on another account you must install the suite passphrase (12.6.1), and configure passwordless ssh so that the port number can be retrieved from the remote port file. Then use the --user and --host command options to connect:
If you know the port number of the target suite, give it on the command line to prevent the port-retrieving ssh connection being attempted:
Possession of a suite passphrase gives full control over the suite, and ssh access to the port file also implies full access to the suite host account, so it is recommended that this only be used to interact with your own suites running on other hosts. We plan to implement finer-grained authentication in the future to allow suite owners to grant read-only access to others.
A connection timeout can be set in site and user global config files (see 6) so that messaging commands cannot hang indefinitely if the suite is not responding (this can be caused by suspending a suite with Ctrl-Z) thereby preventing the task from completing. The same can be done on the command line for other suite-connecting user commands, with the --pyro-timeout option.
Runahead limiting prevents the fastest tasks in a suite from getting too far ahead of the slowest ones. Newly spawned tasks are released to the task pool only when they fall below the runahead limit. A low runhead limit can prevent cylc from interleaving cycles, but it will not stall a suite unless it fails to extend out past a future trigger (see 9.3.4.10). A high runahead limit may allow fast tasks that are not constrained by dependencies or clock-triggers to spawn far ahead of the pack, which could have performance implications for the suite daemon when running very large suites. Succeeded and failed tasks are ignored when computing the runahead limit.
The preferred runahead limiting mechanism restricts the number of consecutive active cycle points. The default value is three active cycle points; see A.3.5. Alternatively the interval between the slowest and fastest tasks can be specified as hard limit; see A.3.4.
Large suites can potentially overwhelm task hosts by submitting too many tasks at once. You can prevent this with internal queues, which limit the number of tasks that can be active (submitted or running) at the some time.
A queue is defined by a name; a limit, which is the maximum number of active tasks allowed for the queue; and a list of members, assigned by task or family name.
Queue configuration is done under the [scheduling] section of the suite.rc file (like dependencies, internal queues constrain when a task runs).
By default every task is assigned to the default queue, which by default has a zero limit (interpreted by cylc as no limit). To use a single queue for the whole suite just set the default queue limit:
To use additional queues just name each one, set their limits, and assign members:
Any tasks not assigned to a particular queue will remain in the default queue. The queues example suite illustrates how queues work by running two task trees side by side (as seen in the graph GUI) each limited to 2 and 3 tasks respectively:
title = demonstrates internal queueing
description = """
Two trees of tasks: the first uses the default queue set to a limit of
two active tasks at once; the second uses another queue limited to three
active tasks at once. Run via the graph control GUI for a clear view.
"""
[scheduling]
[[queues]]
[[[default]]]
limit = 2
[[[foo]]]
limit = 3
members = n, o, p, fam2, u, v, w, x, y, z
[[dependencies]]
graph = """
a => b & c => fam1:succeed-all => h & i & j & k & l & m
n => o & p => fam2:succeed-all => u & v & w & x & y & z
"""
[runtime]
[[fam1,fam2]]
[[d,e,f,g]]
inherit = fam1
[[q,r,s,t]]
inherit = fam2
See also A.4.1.9 in the Suite.rc Reference.
Tasks can be configured with a list of “retry delay” periods, as ISO 8601 durations, such that if a task fails it will go into a temporary retrying state and then automatically resubmit itself after the next specified delay period expires. A usage example is shown in the suite listed below under 12.15.
See also A.2.8 and A.4.1.20 in the Suite.rc Reference.
Cylc can call nominated event handlers when certain suite or task events occur. This is intended to facilitate centralized alerting and automated handling of critical events. Event handlers can send an email or an SMS, call a pager, and so on; or intervene in the operation of their own suite using cylc commands. cylc [hook]email-suite and cylc [hook]email-task are example event handlers packaged with cylc.
Event handlers can be located in the suite bin directory, otherwise it is up to you to ensure their location is in $PATH (in the shell in which cylc runs, on the suite host).
Task event handlers are passed the following arguments by the suite daemon:
where EVENT is one of the following:
MESSAGE, if provided, describes what has happened, and TASKID identifies the task (NAME.CYCLE for cycling tasks).
The retry event occurs if a task fails and has any remaining retries configured (see 12.14). The event handler will be called as soon as the task fails, not after the retry delay period when it is resubmitted.
Note that event handlers are called by cylc itself, not by the running tasks so if you wish to pass them additional information via the environment you must use [cylc] → [[environment]], not task runtime environments.
Here is an example suite that tests the retry and failed events. The handler in this case simply echoes its command line arguments to suite stdout.
The cylc reload command reloads the suite definition at run time. This allows: (a) changing task config such as command scripting or environment; (b) adding tasks to, or removing them from, the suite definition, at run time - without shutting the suite down and restarting it. (It is easy to shut down and restart cylc suites, but reloading may be useful if you don’t want to wait for long-running tasks to finish first).
Note that defined tasks can be already be added to or removed from a running suite with the ’cylc insert’ and ’cylc remove’ commands; the reload command allows addition and removal of task definitions. If a new task is definition is added (and used in the graph) you will still need to manually insert an instance of it (with a particular cycle point) into the running suite. If a task definition (and its use in the graph) is deleted, existing task proxies of the of the deleted type will run their course after the reload but new instances will not be spawned. Changes to a task definition will only take effect when the next task instance is spawned (existing instances will not be affected).
Some HPC facilities allow job preemption: the resource manager can kill or suspend running low priority jobs in order to make way for high priority jobs. The preempted jobs may then be automatically restarted by the resource manager, from the same point (if suspended) or requeued to run again from the start (if killed). If a running cylc task gets suspended or hard-killed (kill -9 <PID> is not a trappable signal so cylc cannot detect task failure in this case) and then later restarted, it will just appear to cylc as if it takes longer than normal to run. If the job is soft-killed the signal will be trapped by the task job script and a failure message sent, resulting in cylc putting the task into the failed state. When the preempted task restarts and sends its started message cylc would normally treat this as an error condition (a dead task is not supposed to be sending messages) - a warning will be logged and the task will remain in the failed state. However, if you know that preemption is possible on your system you can tell cylc that affected tasks should be resurrected from the dead, to carry on as normal if progress messages start coming in again after a failure:
To test this in any suite, manually kill a running task then, after cylc registers the task failed, resubmit the killed job manually by cutting-and-pasting the original job submission command from the suite stdout stream.
The cylc broadcast command overrides [runtime] settings in a running suite. This can be used to communicate information to downstream tasks by broadcasting environment variables (communication of information from one task to another normally takes place via the filesystem, i.e. the input/output file relationships embodied in inter-task dependencies). Variables (and any other runtime settings) may be broadcast to all subsequent tasks, or targeted specifically at a specific task, all subsequent tasks with a given name, or all tasks with a given cycle point; see broadcast command help for details.
Broadcast settings targeted at a specific task ID or cycle point expire and are forgotten as the suite moves on. Un-targeted variables and those targetted at a task name persist throughout the suite run, even across restarts, unless manually cleared using the broadcast command - and so should be used sparingly.
When a suite is started with the cylc run command (cold or warm start) the cycle point at which it starts can be given on the command line or hardwired into the suite.rc file:
or,
An initial cycle given on the command line will override one in the suite.rc file.
In the case of a cold start only the initial cycle point is passed through to task execution environments as $CYLC_SUITE_INITIAL_CYCLE_POINT. The value is then stored in suite state dumps and persists across restarts, but it does get wiped out (set to None) after a warm start, because a warm start is really an implicit restart in which all state information is lost (except that the previous cycle is assumed to have completed).
The $CYLC_SUITE_INITIAL_CYCLE_POINT variable allows tasks to determine if they are running in the initial cold-start cycle point, when different behaviour may be required, or in a normal mid-run cycle point. Note however that an initial R1 graph section is now the preferred way to get different behaviour at suite start-up.
Since cylc-4.6.0 any cylc suite can run in live, simulation, or dummy mode. Prior to that release simulation mode was a hybrid mode that replaced real tasks with local dummy tasks. This allowed local simulation testing of any suite, to get the scheduling right without running real tasks, but running dummy tasks locally does not add much value over a pure simulation (in which no tasks are submitted at all) because all job submission configuration has to be ignored and most task job script sections have to be cut out to avoid any code that could potentially be specific to the intended task host. So at 4.6.0 we replaced this with a pure simulation mode (task proxies go through the running state automatically within cylc, and no dummy tasks are submitted to run) and a new dummy mode in which only the real task command scripting is dummied out - each dummy task is submitted exactly as the task it represents on the correct host and in the same execution environment. A successful dummy run confirms not only that the scheduling works correctly but also tests real job submission, communication from remote task hosts, and the real task job scripts (in which errors such as use of undefined variables will cause a task to fail).
The run mode, which defaults to live, is set on the command line (for run and restart):
but you can configure the suite to force a particular run mode,
This can be used, for example, for demo suites that necessarily run out of their original context; or to temporarily prevent accidental execution of expensive real tasks during suite development.
Dummy mode task command scripting just prints a message and sleeps for ten seconds by default, but you can override this behaviour for particular tasks or task groups if you like. Here’s how to make a task sleep for twenty seconds and then fail in dummy mode:
Finally, in simulation mode each task takes between 1 and 15 seconds to “run” by default, but you can also alter this for particular tasks or groups of tasks:
Note that to get a failed simulation or dummy mode task to succeed on re-triggering, just change the suite.rc file appropriately and reload the suite definition at run time with cylc reload SUITE before re-triggering the task.
Dummy mode is equivalent to commenting out each task’s command scripting to expose the default scripting.
The run mode is recorded in the suite state dump file. Cylc will not let you restart a non-live mode suite in live mode, or vice versa - any attempt to do the former would certainly be a mistake (because the simulation mode dummy tasks do not generate any of the real outputs depended on by downstream live tasks), and the latter, while feasible, would corrupt the live state dump by turning it over to simulation mode. The easiest way to test a live suite in simulation mode, if you don’t want to obliterate the current state dump by doing a cold or warm start (as opposed to a restart from the previous state) is to take a quick copy of the suite and run the copy in simulation mode. However, if you really want to run a live suite forward in simulation mode without copying it, do this:
Reference tests are finite-duration suite runs that abort with non-zero exit status if any of the following conditions occur (by default):
The default shutdown event handler for reference tests is cylc hook check-triggering which compares task triggering information (what triggers off what at run time) in the test run suite log to that from an earlier reference run, disregarding the timing and order of events - which can vary according to the external queueing conditions, runahead limit, and so on.
To prepare a reference log for a suite, run it with the --reference-log option, and manually verify the correctness of the reference run.
To reference test a suite, just run it (in dummy mode for the most comprehensive test without running real tasks) with the --reference-test option.
A battery of automated reference tests is used to test cylc before posting a new release version. Reference tests can also be used at cylc upgrade time to check that the upgrade will not break your own complex suites - the triggering check will catch any bug that causes a task to run when it shouldn’t, for instance; even in a dummy mode reference test the full task job script (sans real command scripting) has to execute successfully on the proper task host by the proper job submission method.
Reference tests can be configured with the following settings:
If the default reference test is not sufficient for your needs, firstly note that you can override the default shutdown event handler, and secondly that the --reference-test option is merely a short cut to the following suite.rc settings which can also be set manually if you wish:
The cylc suite-state command, which interrogates suite run databases, has a polling mode that waits on a given task achieving a given state. See cylc suite-state --help for command options and defaults.
The suite graph notation also allows you to define local tasks that, in effect, represent tasks in other suites by automatically polling for them using the cylc suite-state command. Here’s how to trigger a task bar off a task foo in another suite called other.suite:
Local task FOO will poll for the success of foo in suite other.suite at the same cycle point. Other task states can be polled like this,
Default polling parameters (the maximum number of polls and the interval between them) are printed by cylc suite-state --help. These can be configured if necessary under the local polling task runtime section:
The remote suite does not have to be running when polling commences (or at all if the remote condition has already been achieved) because the command interrogates the suite run database, not the suite server process.
For suites owned by others or those with run databases in non-standard locations use the --run-dir option or, in-suite,
To trigger off remote tasks with different cycle points just arrange for the local polling task to be on the same cycling sequence as the remote task that it represents. For instance, if local task cat cycles 6-hourly at 0,6,12,18 but needs to trigger off a remote task dog with cycle points of 3,9,15,21 hours,
This results in DOG having cycle points of 3,9,15,21 - the sames as dog in other.suite.
The following topics have yet to be documented in detail.
Small groups of cylc users can of course share suites by manual copying, and generic revision control tools can be used on cylc suites as for any collection of files. Beyond this cylc does not have a built-in solution for suite storage and discovery, revision control, and deployment, on a network. That is not cylc’s core purpose, and large sites may have preferred revision control systems and suite meta-data requirements that are difficult to anticipate. We can, however, recommend the use of Rose to do all of this very easily and elegantly with cylc suites.
Rose is a framework for managing and running suites of scientific applications, developed at the UK Met Office for use with cylc. It is available under the open source GPL license.
A suite can contain a small number of large, internally complex tasks; a large number of small, simple tasks; or anything in between. Cylc can easily handle a large number of tasks, however, so there are definite advantages to fine-graining:
It should be possible to rerun a task by simply resubmitting it for the same cycle point. In other words, failure at any point during execution of a task should not render a rerun impossible by corrupting the state of some internal-use file, or whatever. It is difficult to overstate the usefulness of being able to rerun the same task multiple times, either outside of the suite with cylc submit, or by re-triggering it within the running suite, when debugging a problem.
If a warm-cycled model uses the exact same file names for its restart files regardless of current cycle point, the only cycle point that can subsequently run successfully is the next one. Instead, restart files should be labelled with current cycle point and maintained in a simple rolling archive. Then you can easily rerun the task for any cycle point still in the archive.
If a task does not depend on files generated by another task then generally speaking it should not trigger off that task in the suite scheduling graph. Unnecessary dependence between tasks restricts functional parallelism at run time, and it makes the suite more difficult to understand. If you need to restrict the number of tasks that are active at once, use runahead limiting (12.12) and internal queues (12.13).
Putting task cycle point in output file or directory names makes archiving and cleanup easier, and it facilitates re-runnability by ensuring that important files do not get overwritten from one cycle to the next.
The cylc cycle-point command computes offsets from a given or current cycle point, and can insert the resulting computed date-time into a filename template string.
Dependence between tasks usually, although not always, take the form of files generated by one task and used by others. It is possible to manage these files across a suite without compromising suite flexibility and portability with hard wired I/O locations.
You may be able to have all tasks, or groups of tasks that need to cooperate, read and write from a common workspace, thereby avoiding the need to explicitly move files around. The suite share directory ($CYLC_SUITE_SHARE_DIR) is provided for this purpose. Similarly, task work directories are private to each task by default but they can be shared to allow multiple tasks to simply read and write from their current working directory. Even if you use other custom I/O directories, define their locations in the suite.rc file rather than hard wiring them into task implementation. Shared workspace locations can be passed to tasks as needed without modifying the task implementation, like this:
Special tasks can be used to move files around, from one task’s output directory to another’s input directory. This should only be necessary across host or filesystem boundaries, however; otherwise simply reference shared locations as shown above.
If your suite contains multiple logically distinct tasks that have similar functionality (e.g. tasks that move files around or generate similar products from different datasets) just have them all call the same underlying command, script, or executable, but provide different input parameters as required.
If all I/O is automatically done in suite-specific locations, such as the under suite share and work directories ($CYLC_SUITE_SHARE_DIRECTORY and $CYLC_TASK_WORK_DIRECTORY), you should be able to run multiple copies of the same suite without interference between them, and other users should be able to copy and run your suites with minimal modification.
Where possible a task should not rely on the action of another task, except for the inputs embodied in the suite dependency graph that it has no choice but to depend on. This makes it as easy as possible to run single tasks alone during suite development and debugging. For example, tasks should create their own output directories if necessary rather than assuming their existence due to the action of another task. Note that if the existing task implementation does not handle output directory creation you can do it in suite pre-command scripting or similar.
Tasks can (of course) run external commands, scripts, and executables; and they can read or otherwise make use of external files. In some cases this may be necessary, but it does leave suites vulnerable to external breakages. Alternatively, suites can be more or less completely self-contained (aside from exposure to network, filesystem, and OS problems) if they have private copies of every file they need at run time. Tasks can access files stored under their suite definition directory via $CYLC_SUITE_DEF_PATH, and the suite bin directory is automatically added to $PATH in the task execution environment. If you have multiple suites there may be a tradeoff between self-containment and duplication of files, but this does not particularly matter if you can automatically extract, build, and install suite files from external repositories prior to, or at the start of, a suite run.
A suite definition and any files stored with it should be version controlled, and a particular revision extracted before a run. The extracted source suite will be a repository clone or working copy, depending on your choice of revision control software, and can be used for further development. The source files should then be installed to another location where the suite will actually be executed (the cylc suite run directory is ideal for this). External files may also be installed into the suite at this time, prior to the run, or by special deployment tasks that run at suite start-up. This makes self-containment easier to achieve, and the clean separation of source and installed suite allows further development without breaking a running suite. Rose (14.1) supports this mode of working with cylc suites.
Correct scheduling is not necessarily equivalent to orderly generation of products in strict date-time order. Under cylc a product generation task will trigger as soon as its private prerequisites are satisfied regardless of whether other tasks at the same cycle point have finished or have yet to run. If your product presentation system demands that all products are uploaded in order, then be aware that this may be quite inefficient if your suite ever has to catch up from a delay or run over historical data, but if necessary you can force tasks to run in the right order even if their true dependencies do not require that. One way to do this is to declare the product upload task to be sequential, which is equivalent to making it depend on its own previous instance (see 9.3.4.12).
Most tasks submit as soon as their prerequisites (task triggers) are satisfied, but clock-triggered tasks also wait on a wall clock time expressed as an offset from their cycle point. For example, task foo below will trigger (other dependencies allowing) 2 hours after the wall clock time passes its cycle point (which in this case must be a date-time).
Tasks that wait on external real time data should have a clock-trigger to delay submission until roughly the expected time of data availability, otherwise they may clutter up batch scheduler queues by submitting hours earlier. Similarly, suite polling tasks (for inter-suite dependence in real time operation) should use a clock-trigger to delay their submission until the expected time of the remote suite event.
Some tasks wait on external events and therefore need to repeatedly check and wait for the event before reporting eventual success (or perhaps failure after a timeout). For example, a task that waits for a file to appear on an ftp server. Typically these should be clock-triggered tasks (see above), but once triggered there are two ways to handle the repeated checking: the task itself could implement a check-and-wait loop; or you could just configure multiple retries for the task, in the suite definition (see 12.14).
Cylc suites, without modification, can handle real time and delayed operation equally well. In caught-up real time operation, clock-triggered tasks constrain the behaviour of the whole suite, or at least of any tasks downstream of them in the dependency graph. In delayed or historical operation clock-triggered tasks will not constrain the suite at all, and cylc’s cycle point interleaving abilities come to the fore, because the clock-trigger times have already passed. But if a clock-triggered task catches up to the wall clock, it will automatically wait again. In this way cylc suites naturally transition between delayed and real time operation as required.
To help avoid suite maintenance errors in the future, properties shared by multiple tasks (job submission settings, environment variables, command scripting, etc.) should be defined only once, using runtime inheritance (9.4) or Jinja2 variables (9.6).
Multiple inheritance is efficient when tasks share many properties, but Jinja2 variables may be preferred when a small number of properties are shared by tasks that don’t have anything else in common (e.g. a single environment variable for the location of a shared file).
For environment variables in particular it may be tempting to define all variables for all tasks once under [root], but this is analagous to overuse of global variables in programming and it can make it difficult to determine which variables matter to which tasks. Environment filters (A.4.1.22) can be used to make this safer, but generally it is best to provide each task with only the variables that it needs. It is difficult to be sure if a task really needs a variable that is passed to it, but you can be sure that it does not use a variable that is not passed to it.
Finally, Jinja2 can also be used to avoid polluting task environments with variables used for the sole purpose of deriving other variables at task run time. Instead of this:
do this:
If the values of these Jinja2 variables are needed in external scripts, just translate them directly in environment sections:
If you find yourself writing runtime scripting to change a task’s behaviour in some cycle points, consider that the graph is usually the proper place to express this sort of thing. Use different task names, but have them inherit common properties to avoid duplication. Instead of this:
do this:
Similarly, if your task has a different behaviour at the initial or final cycle point, consider using an R1 syntax to separate out the functionality.
Effective visualization can make complex suites easier to understand. Collapsible task families for visualization are defined by the first parents in the runtime namespace hierarchy. Tasks should generally be grouped into visualization families that reflect their purpose within the structure of the suite rather than technical detail such as common job submission method or task host. This often coincides nicely with common configuration inheritance requirements, but if it doesn’t you can use an empty namespace as a first parent for visualization:
and you can demote parents from primary to secondary:
Good style is to some extent a matter of taste. That said, for collaborative development of complex systems it is important to settle on a clear and consistent style, and you may find the following suggestions useful. Note that the boundary between this section (style) and the previous (design) is somewhat arbitrary.
The suite.rc file format consists of item = value pairs under nested section headings. Clear indentation is the best way to show local nesting level inside large blocks.
Don’t align item = value pairs on the = character - this does not show nesting level clearly and it pushes everything off to the right:
The following layout does preserve proper indentation on the left, but the whole block may need reformatting after changing one line, which pollutes your revision history with spurious changes:
is preferred over this (or similar):
The extra whitespace here translates directly to spurious indentation in the task job script. As it happens this is just an aesthetic problem in bash scripts, but for Python job scripts (which cylc may support in the future) it would be a technical error.
Comments should be minimal, but not too minimal. If context and clear item names will do, leave it at that. Extremely verbose comments tend to be neglected and eventually get out of sync with the code, a result that may be worse than having no comments at all.
Keep to the standard maximum line length of 79 characters where possible. Very long lines affect readability, may pose a problem for auto-line-breaking in text editors, and make side-by-side diff display less effective.
Graph lines can also be split up without line breaks, like this:
Use UPPERCASE_NAMES for families and lowercase_names for tasks, so that you can tell which is which at a glance.
Trivial task scripting may be inlined in the suite definition but anything more should be written to a script file. This keeps the suite definition tidy, it allows proper shell-mode text editing, and it allows separate command line testing of the script during development or debugging.
This appendix defines all legal suite definition config items. Embedded Jinja2 code (see 9.6) must process to a valid raw suite.rc file. See also 9.2 for a descriptive overview of suite.rc files, including syntax (9.2.1).
The only top level configuration items at present are the suite title and description.
A single line description of the suite. It is displayed in the db viewer window and can be retrieved at run time with the cylc show command.
A multi-line description of the suite. It can be retrieved by the db viewer right-click menu, or at run time with the cylc show command.
This section is for configuration that is not specifically task-related.
If this item is set cylc will abort if the suite is not started in the specified mode. This can be used for demo suites that have to be run in simulation mode, for example, because they have been taken out of their normal operational context; or to prevent accidental submission of expensive real tasks during suite development.
Cylc runs off the suite host’s system clock by default. This item allows you to run the suite in UTC even if the system clock is set to local time. Clock-triggered tasks will trigger when the current UTC time is equal to their cycle point date-time plus offset; other time values used, reported, or logged by the suite daemon will usually also be in UTC.
To just alter the timezone used in the date/time cycle point format, see A.2.5. To just alter the number of expanded year digits (for years below 0 or above 9999), see A.2.4.
Cylc usually uses a CCYYMMDDThhmmZ (Z in the special case of UTC) or CCYYMMDDThhmm+hhmm format (+ standing for + or - here) for writing down date/time cycle points, which follows one of the basic formats outlined in the ISO 8601 standard. For example, a cycle point on the 3rd of February 2001 at 4:50 in the morning, UTC (+0000 timezone), would be written 20010203T0450Z. Similarly, for the the 3rd of February 2001 at 4:50 in the morning, +1300 timezone, cylc would write 20010203T0450+1300.
You may use the isodatetime library’s syntax to write dates and times in ISO 8601 formats - CC for century, YY for decade and decadal year, +X for expanded year digits and their positive or negative sign, thereafter following the ISO 8601 standard example notation except for fractional digits, which are represented as ,ii for hh, ,nn for mm, etc. For example, to write date/times as week dates with fractional hours, set cycle point format to CCYYWwwDThh,iiZ e.g. 1987W041T08,5Z for 08:30 UTC on Monday on the fourth ISO week of 1987.
You can also use a subset of the strptime/strftime POSIX standard - supported tokens are \%F, \%H, \%M, \%S, \%Y, \%d, \%j, \%m, \%s, \%z.
To use the old cylc date-time format (e.g. 2014020106 for 06:00 on the 1st of February 2014), set cycle point format to \%Y\%m\%d\%H.
Please note that using characters like ”/” is not allowed, as it will break task output files (”/” is a reserved character in POSIX file and directory naming). Using ”:” is also not allowed, as it is likely to interfere with usage of commands like ”rsync” when applied to task output files.
For years below 0 or above 9999, the ISO 8601 standard specifies that an extra number of year digits and a sign should be used. This extra number needs to be written down somewhere (here).
For example, if this extra number is set to 2, 00Z on the 1st of January in the year 10040 will be represented as +0100400101T0000Z (2 extra year digits used). With this number set to 3, 06Z on the 4th of May 1985 would be written as +00019850504T0600Z.
This number defaults to 0 (no sign or extra digits used).
If you set UTC mode to True (A.2.2) then this will default to Z. If you use a custom cycle point format (A.2.3), you should specify the timezone choice (or null timezone choice) here as well.
You may set your own time zone choice here, which will be used for all date/time cycle point dumping. Time zones should be expressed as ISO 8601 time zone offsets from UTC, such as +13, +1300, -0500 or +0645, with Z representing the special +0000 case. Cycle points will be converted to the time zone you give and will be represented with this string at the end.
Cycle points that are input without time zones (e.g. as an initial cycle point setting) will use this time zone if set. If this isn’t set (and UTC mode is also not set), then they will default to the current local time zone.
Note that the ISO standard also allows writing the hour and minute separated by a ”:” (e.g. +13:00) - however, this is not recommended, given that the time zone is used as part of task output filenames.
Cylc does not normally abort if tasks fail, but if this item is turned on it will abort with exit status 1 if any task fails.
If this is turned on cylc will write the resolved dependencies of each task to the suite log as it becomes ready to run (a list of the IDs of the tasks that actually satisfied its prerequisites at run time). Mainly used for cylc testing and development.
Cylc has internal “hooks” to which you can attach handlers that are called by the suite daemon whenever certain events occur. This section configures suite event hooks; see A.4.1.20 for task event hooks.
Event handlers can send an email or an SMS, call a pager, intervene in the operation of their own suite, or whatever. They can be held in the suite bin directory, otherwise it is up to you to ensure their location is in $PATH (in the shell in which cylc runs, on the suite host). cylc [hook] email-suite is a simple suite event handler.
Suite event handlers are called by the suite daemon with the following arguments:
where,
Additional information can be passed to event handlers via [cylc] → [[environment]].
[cylc] → [[event hooks]] → EVENT handler A list of one or more event handlers to call when one of the following EVENTs occurs:
Item details:
[cylc] → [[event hooks]] → timeout If a timeout is set and the timeout event is handled, the timeout event handler(s) will be called if the suite times out before it finishes. The timer is set initially at suite start up.
[cylc] → [[event hooks]] → reset timer If True (the default) the suite timer will continually reset after any task changes state, so you can time out after some interval since the last activity occured rather than on absolute suite execution time.
[cylc] → [[event hooks]] → abort on timeout If a suite timer is set (above) this will cause the suite to abort with error status if the suite times out while still running.
[cylc] → [[event hooks]] → abort if EVENT handler fails Cylc does not normally care whether an event handler succeeds or fails, but if this is turned on the EVENT handler will be executed in the foreground (which will block the suite while it is running) and the suite will abort if the handler fails.
Environment variables defined in this section are passed to suite and task event handlers.
[cylc] → [[environment]] → __VARIABLE__ Replace __VARIABLE__ with any number of environment variable assignment expressions. Values may refer to other local environment variables (order of definition is preserved) and are not evaluated or manipulated by cylc, so any variable assignment expression that is legal in the shell in which cylc is running can be used (but see the warning above on variable expansions, which will not be evaluated). White space around the ‘=’ is allowed (as far as cylc’s file parser is concerned these are just suite configuration items).
Reference tests are finite-duration suite runs that abort with non-zero exit status if cylc fails, if any task fails, if the suite times out, or if a shutdown event handler that (by default) compares the test run with a reference run reports failure. See 12.21.
[cylc] → [[reference test]] → suite shutdown event handler A shutdown event handler that should compare the test run with the reference run, exiting with zero exit status only if the test run verifies.
As for any event handler, the full path can be ommited if the script is located somewhere in $PATH or in the suite bin directory.
[cylc] → [[reference test]] → required run mode If your reference test is only valid for a particular run mode, this setting will cause cylc to abort if a reference test is attempted in another run mode.
[cylc] → [[reference test]] → allow task failures A reference test run will abort immediately if any task fails, unless this item is set, or a list of expected task failures is provided (below).
[cylc] → [[reference test]] → expected task failures A reference test run will abort immediately if any task fails, unless allow task failures is set (above) or the failed task is found in a list IDs of tasks that are expected to fail.
[cylc] → [[reference test]] → live mode suite timeout The timeout value, expressed as an ISO 8601 duration/interval, after which the test run should be aborted if it has not finished, in live mode. Test runs cannot be done in live mode unless you define a value for this item, because it is not possible to arrive at a sensible default for all suites.
[cylc] → [[reference test]] → simulation mode suite timeout The timeout value in minutes after which the test run should be aborted if it has not finished, in simulation mode. Test runs cannot be done in simulation mode unless you define a value for this item, because it is not possible to arrive at a sensible default for all suites.
[cylc] → [[reference test]] → dummy mode suite timeout The timeout value, expressed as an ISO 8601 duration/interval, after which the test run should be aborted if it has not finished, in dummy mode. Test runs cannot be done in dummy mode unless you define a value for this item, because it is not possible to arrive at a sensible default for all suites.
This section allows cylc to determine when tasks are ready to run.
Cylc runs using the proleptic Gregorian calendar by default. This item allows you to either run the suite using the 360 day calendar (12 months of 30 days in a year) or using integer cycling.
In a cold start each cycling task (unless specifically excluded under [special tasks]) will be loaded into the suite with this cycle point, or with the closest subsequent valid cycle point for the task. Note that special cold-start tasks are not loaded in a warm start. If this item is provided you can override it on the command line or in the gcylc suite start panel.
In date-time cycling, if you do not provide time zone information for this, it will be assumed to be local time, or in UTC if A.2.2 is set, or in the time zone determined by A.2.5 if that is set.
Cycling tasks are held once they pass the final cycle point, if one is specified. Once all tasks have achieved this state the suite will shut down. If this item is provided you can override it on the command line or in the gcylc suite start panel.
In date-time cycling, if you do not provide time zone information for this, it will be assumed to be local time, or in UTC if A.2.2 is set, or in the A.2.5 if that is set.
Runahead limiting prevents the fastest tasks in a suite from getting too far ahead of the slowest ones, as documented in 12.12.
This config item specifies a hard limit as a cycle interval between the slowest and fastest tasks. It is deprecated in favour of the newer default limiting by max active cycle points (A.3.5).
Runahead limiting prevents the fastest tasks in a suite from getting too far ahead of the slowest ones, as documented in 12.12.
This config item supersedes the deprecated hard runahead limit (A.3.4). It allows up to N (default 3) consecutive cycle points to be active at any time, adjusted up if necessary for any future triggering.
Configuration of internal queues, by which the number of simultaneously active tasks (submitted or running) can be limited, per queue. By default a single queue called default is defined, with all tasks assigned to it and no limit. To use a single queue for the whole suite just set the limit on the default queue as required. See also 12.13.
[scheduling] → [[queues]] → [[[__QUEUE__]]] Section heading for configuration of a single queue. Replace __QUEUE__ with a queue name, and repeat the section as required.
[scheduling] → [[queues]] → [[[__QUEUE__]]] → limit The maximum number of active tasks allowed at any one time, for this queue.
[scheduling] → [[queues]] → [[[__QUEUE__]]] → members A list of member tasks, or task family names, to assign to this queue (assigned tasks will automatically be removed from the default queue).
This section is used to identify any tasks with several kinds of special behaviour. Family names can be used in special task lists as shorthand for listing all member tasks.
[scheduling] → [[special tasks]] → clock-triggered Clock-triggered tasks wait on a wall clock time specified as an offset from their own cycle point, in addition to dependence on other tasks. Clock-triggers can be used to make tasks that wait on external real time data trigger at the expected time of data availability, or to make suite polling tasks trigger at the expected time of the remote suite event. In delayed or historical operation clock-triggered tasks do not constrain the suite until they catch up to the wall clock.
[scheduling] → [[special tasks]] → sequential Sequential tasks are automatically given dependence on their own predecessor. This is equivalent to use of explicit inter-cycle triggers in the graph, except that the automatic version does not show in suite graph visualization. For more on sequential tasks see 9.3.4.12 and 15.4.
[scheduling] → [[special tasks]] → cold-start A cold-start task is one-off task used to satisfy the dependence of an associated task with the same cycle point, on outputs from a previous cycle - when those outputs are not available. The primary use for this is to cold-start a warm-cycled forecast model that normally depends on restart files (e.g. model background fields) generated by its previous forecast, when there is no previous forecast. This is required when cold-starting the suite, but cold-start tasks can also be inserted into a running suite to restart a model that has had to skip some cycles after running into problems. Cold-start tasks can invoke real cold-start processes, or they can just be dummy tasks that represent some external process that has to be completed before the suite is started. Unlike start-up tasks, dependence on cold-start tasks is preseverved in subsequent cycles so they must typically be used in OR’d conditional expressions to avoid holding up the suite.
[scheduling] → [[special tasks]] → one-off Synchronous one-off tasks have an associated cycle point but do not spawn a successor. Synchronous start-up and cold-start tasks are automatically one-off tasks and do not need to be listed here. Dependence on one-off tasks is not restricted to the first cycle.
[scheduling] → [[special tasks]] → exclude at start-up Any task listed here will be excluded from the initial task pool (this goes for suite restarts too). If an inclusion list is also specified, the initial pool will contain only included tasks that have not been excluded. Excluded tasks can still be inserted at run time. Other tasks may still depend on excluded tasks if they have not been removed from the suite dependency graph, in which case some manual triggering, or insertion of excluded tasks, may be required.
[scheduling] → [[special tasks]] → include at start-up If this list is not empty, any task not listed in it will be excluded from the initial task pool (this goes for suite restarts too). If an exclusion list is also specified, the initial pool will contain only included tasks that have not been excluded. Excluded tasks can still be inserted at run time. Other tasks may still depend on excluded tasks if they have not been removed from the suite dependency graph, in which case some manual triggering, or insertion of excluded tasks, may be required.
The suite dependency graph is defined under this section. You can plot the dependency graph as you work on it, with cylc graph or by right clicking on the suite in the db viewer. See also 9.3.
[scheduling] → [[dependencies]] → graph The dependency graph for a completely non-cycling suites can go here. See also A.3.8.2.1 below and 9.3, for graph string syntax.
[scheduling] → [[dependencies]] → [[[__RECURRENCE__]]] __RECURRENCE__ section headings define the sequence of cycle points for which the subsequent graph section is valid. These should be specified in our ISO 8601 derived sequence syntax, or similar for integer cycling:
See 9.3.3 for more on recurrence expressions, and how multiple graph sections combine.
[scheduling] → [[dependencies]] → [[[__RECURRENCE__]]] → graph The dependency graph for a given recurrence section goes here. Syntax examples follow; see also 9.3 and 9.3.4.
This section is used to specify how, where, and what to execute when tasks are ready to run. Common configuration can be factored out in a multiple-inheritance hierarchy of runtime namespaces that culminates in the tasks of the suite. Order of precedence is determined by the C3 linearization algorithm as used to find the method resolution order in Python language class hiearchies. For details and examples see 9.4.
Replace __NAME__ with a namespace name, or a comma separated list of names, and repeat as needed to define all tasks in the suite. Names may contain letters, digits, underscores, and hyphens. A namespace represents a group or family of tasks if other namespaces inherit from it, or a task if no others inherit from it.
If multiple names are listed the subsequent settings apply to each.
All namespaces inherit initially from root, which can be explicitly configured to provide or override default settings for all tasks in the suite.
[runtime] → [[__NAME__]] → inherit A list of the immediate parent(s) this namespace inherits from. If no parents are listed root is assumed.
[runtime] → [[__NAME__]] → title A single line description of this namespace. It is displayed by the cylc list command and can be retrieved from running tasks with the cylc show command.
[runtime] → [[__NAME__]] → description A multi-line description of this namespace, retrievable from running tasks with the cylc show command.
[runtime] → [[__NAME__]] → initial scripting Initial scripting is executed at the top of the task job script just before the cylc task started message call is made, and before the task execution environment is configured - so it does not have access to any suite or task environment variables. The original intention was to allow remote tasks to source login scripts before calling the first cylc command, e.g. to set $PYTHONPATH if Pyro has been installed locally. Note however that the remote task invocation mechanism now automatically sources both /etc/profile and $HOME/.profile if they exist. For other uses pre-command scripting should be used if possible because it can has access to the task execution environment.
[runtime] → [[__NAME__]] → environment scripting Environment scripting is inserted into the task job script between the cylc-defined environment (suite and task identity, etc.) and the user-defined task runtime environment - i.e. it has access to the cylc environment, and the task environment has access to the results of this scripting.
[runtime] → [[__NAME__]] → command scripting The scripting to execute when the associated task is ready to run - this can be a single command or multiple lines of scripting.
[runtime] → [[__NAME__]] → pre-command scripting Scripting to be executed immediately before the command scripting. This would typically be used to add scripting to every task in a family (for individual tasks you could just incorporate the extra commands into the main command scripting). See also post-command scripting, below.
[runtime] → [[__NAME__]] → post-command scripting Scripting to be executed immediately after the command scripting. This would typically be used to add scripting to every task in a family (for individual tasks you could just incorporate the extra commands into the main command scripting). See also pre-command scripting, above.
[runtime] → [[__NAME__]] → retry delays A list of ISO 8601 time duration/intervals after which to resubmit the task if it fails. The variable $CYLC_TASK_TRY_NUMBER in the task execution environment is incremented each time, starting from 1 for the first try - this can be used to vary task behavior by try number.
[runtime] → [[__NAME__]] → submission polling intervals A list of intervals, expressed as ISO 8601 duration/intervals, with optional multipliers, after which cylc will poll for status while the task is in the submitted state.
For the polling task communications method this overrides the default submission polling interval in the site/user config files (6). For pyro and ssh task communications polling is not done by default but it can still be configured here as a regular check on the health of submitted tasks.
Each list value is used in turn until the last, which is used repeatedly until finished.
Detaching tasks cannot be polled or killed by the suite daemon - see 10.4.
A single interval value is probably appropriate for submission polling.
[runtime] → [[__NAME__]] → execution polling intervals A list of intervals, expressed as ISO 8601 duration/intervals, with optional multipliers, after which cylc will poll for status while the task is in the running state.
For the polling task communications method this overrides the default execution polling interval in the site/user config files (6). For pyro and ssh task communications polling is not done by default but it can still be configured here as a regular check on the health of submitted tasks.
Each list value is used in turn until the last, which is used repeatedly until finished.
Detaching tasks cannot be polled or killed by the suite daemon - see 10.4.
[runtime] → [[__NAME__]] → manual completion If a task’s initiating process detaches and exits before task processing is finished then cylc cannot arrange for the task to automatically signal when it has succeeded or failed. In such cases you must use this configuration item to tell cylc not to arrange for automatic completion messaging, and insert some minimal completion messaging yourself in appropriate places in the task implementation (see 10.4).
[runtime] → [[__NAME__]] → work sub-directory Task command scripting is executed from with automatically created work directories, which can be accessed by their tasks through $CYLC_TASK_WORK_DIR. This items sets the low-level sub-directory name. The default value provides a unique workspace for each task, but this can overridden to make groups of tasks run in the same working directory, thereby providing a share space for tasks that read and write from their current working directories.
[runtime] → [[__NAME__]] → enable resurrection If a message is received from a failed task cylc will normally treat this as an error condition, issue a warning, and leave the task in the “failed” state. But if “enable resurrection” is switched on failed tasks can come back from the dead: if the same task job script is executed again cylc will put the task back into the running state and continue as normal when the started message is received. This can be used to handle HPC-style job preemption wherein a resource manager may kill a running task and reschedule it to run again later, to make way for a job with higher immediate priority. See also 12.17
[runtime] → [[__NAME__]] → [[[dummy mode]]] Dummy mode configuration.
[runtime] → [[__NAME__]] → [[[dummy mode]]] → command scripting The scripting to execute when the associated task is ready to run, in dummy mode - this can be a single command or a multiple lines of scripting.
[runtime] → [[__NAME__]] → [[[dummy mode]]] → disable pre-command scripting This disables pre-command scripting, is likely to contain code specific to the real task, in dummy mode.
[runtime] → [[__NAME__]] → [[[dummy mode]]] → disable post-command scripting This disables post-command scripting, which is likely to contain code specific to the real task, in dummy mode.
[runtime] → [[__NAME__]] → [[[simulation mode]]] Simulation mode configuration.
[runtime] → [[__NAME__]] → [[[simulation mode]]] → run time range This defines a minimum and a maximum duration (expressed as ISO 8601 duration/intervals) which define a range from which the simulation mode task run length will be randomly chosen.
[runtime] → [[__NAME__]] → [[[job submission]]] This section configures the means by which cylc submits task job scripts to run.
[runtime] → [[__NAME__]] → [[[job submission]]] → method See 11 for how job submission works, and how to define new methods. Cylc has a number of built in job submission methods:
[runtime] → [[__NAME__]] → [[[job submission]]] → command template This allows you to override the actual command used by the chosen job submission method. The template’s first %s will be substituted by the job file path. Where applicable the second and third %s will be substituted by the paths to the job stdout and stderr files.
[runtime] → [[__NAME__]] → [[[job submission]]] → shell This is the shell used to interpret the job script submitted by the suite daemon when a task is ready to run. It has no bearing on the shell used in task implementations. Command scripting and suite environment variable assignment expressions must be valid for this shell. The latter is currently hardwired into cylc as export item=value - valid for both bash and ksh because value is entirely user-defined - but cylc would have to be modified slightly to allow use of the C shell.
[runtime] → [[__NAME__]] → [[[job submission]]] → retry delays A list of duration (in ISO 8601 syntax), after which to resubmit if job submission fails.
[runtime] → [[__NAME__]] → [[[remote]]] Configure host and username, for tasks that do not run on the suite host account. Passwordless ssh is used to submit the task by the configured job submission method, so you must distribute your ssh key to allow this. Cylc must be installed on remote task hosts, but of the external software dependencies only Pyro is required there (not even that if ssh messaging is used; see below).
[runtime] → [[__NAME__]] → [[[remote]]] → host The remote host for this namespace. This can be a static hostname, an environment variable that holds a hostname, or a command that prints a hostname to stdout. Host selection commands are executed just prior to job submission. The host (static or dynamic) may have an entry in the cylc site or user config file to specify parameters such as the location of cylc on the remote machine; if not, the corresponding local settings (on the suite host) will be assumed to apply on the remote host.
[runtime] → [[__NAME__]] → [[[remote]]] → owner The username of the task host account. This is (only) used in the passwordless ssh command invoked by the suite daemon to submit the remote task (consequently it may be defined using local environment variables (i.e. the shell in which cylc runs, and [cylc] → [[environment]]).
If you use dynamic host selection and have different usernames on the different selectable hosts, you can configure your $HOME/.ssh/config to handle username translation.
[runtime] → [[__NAME__]] → [[[remote]]] → suite definition directory The path to the suite definition directory on the remote host, needed if remote tasks require access to files stored there (via $CYLC_SUITE_DEF_PATH) or in the suite bin directory (via $PATH). If this item is not defined, the local suite definition directory path will be assumed, with the suite owner’s home directory, if present, replaced by '$HOME' for interpretation on the remote host.
[runtime] → [[__NAME__]] → [[[event hooks]]] Cylc has internal “hooks” to which you can attach handlers that are called by the suite daemon whenever certain events occur. This section configures task event hooks; see A.2.8 for suite event hooks.
Event handlers can send an email or an SMS, call a pager, intervene in the operation of their own suite, or whatever. They can be held in the suite bin directory, otherwise it is up to you to ensure their location is in $PATH (in the shell in which cylc runs, on the suite host). cylc [hook] email-task is a simple task event handler.
Task event handlers are called by the suite daemon with the following arguments:
where,
Additional information can be passed to event handlers via the [cylc] → [[environment]] (but not via task runtime environments - event handlers are not called by tasks).
[runtime] → [[__NAME__]] → [[[event hooks]]] → EVENT handler A list of one or more event handlers to call when one of the following EVENTs occurs:
Item details:
[runtime] → [[__NAME__]] → [[[event hooks]]] → submission timeout If a task has not started after the specified ISO 8601 duration/interval, the submission timeout event handler(s) will be called.
[runtime] → [[__NAME__]] → [[[event hooks]]] → execution timeout If a task has not finished after the specified ISO 8601 duration/interval, the execution timeout event handler(s) will be called.
[runtime] → [[__NAME__]] → [[[event hooks]]] → reset timer If you set an execution timeout the timer can be reset to zero every time a message is received from the running task (which indicates the task is still alive). Otherwise, the task will timeout if it does not finish in the alotted time regardless of incoming messages.
[runtime] → [[__NAME__]] → [[[environment]]] The user defined task execution environment. Variables defined here can refer to cylc suite and task identity variables, which are exported earlier in the task job script, and variable assignment expressions can use cylc utility commands because access to cylc is also configured earlier in the script. See also 9.4.7.
[runtime] → [[__NAME__]] → [[[environment]]] → __VARIABLE__ Replace __VARIABLE__ with any number of environment variable assignment expressions. Order of definition is preserved so values can refer to previously defined variables. Values are passed through to the task job script without evaluation or manipulation by cylc, so any variable assignment expression that is legal in the job submission shell can be used. White space around the ‘=’ is allowed (as far as cylc’s suite.rc parser is concerned these are just normal configuration items).
[runtime] → [[__NAME__]] → [[[environment filter]]] This section contains environment variable inclusion and exclusion lists that can be used to filter the inherited environment. This is not intended as an alternative to a well-designed inheritance hierarchy that provides each task with just the variables it needs. Filters can, however, improve suites with tasks that inherit a lot of environment they don’t need, by making it clear which tasks use which variables. They can optionally be used routinely as explicit “task environment interfaces” too, at some cost to brevity, because they guarantee that variables filtered out of the inherited task environment are not used.
Note that environment filtering is done after inheritance is completely worked out, not at each level on the way, so filter lists in higher-level namespaces only have an effect if they are not overridden by descendants.
[runtime] → [[__NAME__]] → [[[environment filter]]] → include If given, only variables named in this list will be included from the inherited environment, others will be filtered out. Variables may also be explicitly excluded by an exclude list.
[runtime] → [[__NAME__]] → [[[environment filter]]] → exclude Variables named in this list will be filtered out of the inherited environment. Variables may also be implicitly excluded by omission from an include list.
[runtime] → [[__NAME__]] → [[[directives]]] Batch queue scheduler directives. Whether or not these are used depends on the job submission method. For the built-in loadleveler, pbs, and sge methods directives are written to the top of the task job script in the correct format for the method. Specifying directives individually like this allows use of default directives that can be individually overridden at lower levels of the runtime namespace hierarchy.
[runtime] → [[__NAME__]] → [[[directives]]] → __DIRECTIVE__ Replace __DIRECTIVE__ with each directive assignment, e.g. class = parallel
Example directives for the built-in job submission methods are shown in 11.2.
[runtime] → [[__NAME__]] → [[[outputs]]] This section is for registering custom message outputs that other tasks can trigger off instead of the standard triggers. The task implementation must send corresponding messages using the cylc task message command at the appropriate time. See 9.3.4.5 for more information.
[runtime] → [[__NAME__]] → [[[outputs]]] → __OUTPUT__ Replace __OUTPUT__ with one or more labelled output messages, and use the labels in graph trigger notation. Messages should contain a placeholder for the current cycle point ([]) or some offset from it (e.g. [-P2M]).
See 9.3.4.5 for more information.
[runtime] → [[__NAME__]] → [[[suite state polling]]] Configure automatic suite polling tasks as described in 12.22. The items in this section reflect the options and defaults of the cylc suite-state command, except that the target suite name and the --task, --cycle, and --status options are taken from the graph notation.
[runtime] → [[__NAME__]] → [[[suite state polling]]] → run-dir For your own suites the run database location is determined by your site/user config. For other suites, e.g. those owned by others, or mirrored suite databases, use this item to specify the location of the top level cylc run directory (the database should be a suite-name sub-directory of this location).
[runtime] → [[__NAME__]] → [[[suite state polling]]] → interval Polling interval expressed as an ISO 8601 duration/interval.
[runtime] → [[__NAME__]] → [[[suite state polling]]] → max-polls The maximum number of polls before timing out and entering the ‘failed’ state.
[runtime] → [[__NAME__]] → [[[suite state polling]]] → user Username of an account on the suite host to which you have access. The polling cylc suite-state command will be invoked on the remote account.
[runtime] → [[__NAME__]] → [[[suite state polling]]] → host The hostname of the target suite. The polling cylc suite-state command will be invoked on the remote account.
[runtime] → [[__NAME__]] → [[[suite state polling]]] → verbose Run the polling cylc suite-state command in verbose output mode.
Configuration of suite graphing for the cylc graph command (graph extent, styling, and initial family-collapsed state) and the gcylc graph view (initial family-collapsed state). Graphviz documentation of node shapes and so on can be found at http://www.graphviz.org/Documentation.php.
The initial cycle point for graph plotting.
The visualization initial cycle point gets adjusted up if necessary to the suite initial cycling point.
An explicit final cycle point for graph plotting. If used, this overrides the preferred number of cycle points (below).
The visualization final cycle point gets adjusted down if necessary to the suite final cycle point.
The number of cycle points to graph starting from the visualization initial cycle point. This is the preferred way of defining the graph end point, but it can be overridden by an explicit final cycle point (above).
A list of family (namespace) names to be shown in the collapsed state (i.e. the family members will be replaced by a single family node) when the suite is first plotted in the graph viewer or the gcylc graph view. If this item is not set, the default is to collapse all families at first. Interactive GUI controls can then be used to group and ungroup family nodes at will.
Graph edges (dependency arrows) can be plotted in the same color as the upstream node (task or family) to make paths through a complex graph easier to follow.
Graph node labels can be printed in the same color as the node outline.
Set the default attributes (color and style etc.) of graph nodes (tasks and families). Attribute pairs must be quoted to hide the internal = character.
Set the default attributes (color and style etc.) of graph edges (dependency arrows). Attribute pairs must be quoted to hide the internal = character.
If True, the gcylc graph-view write out a dot-language graph file on every change; these can be post-processed into a movie showing how the suite evolves. The frames will be written to the run time graph directory (see below).
Define named groups of graph nodes (tasks and families) which can styled en masse, by name, in [visualization] → [[node attributes]]. Node groups are automatically defined for all task families, including root, so you can style family and member nodes at once by family name.
[visualization] → [[node groups]] → __GROUP__ Replace __GROUP__ with each named group of tasks or families.
Here you can assign graph node attributes to specific nodes, or to all members of named groups defined in [visualization] → [[node groups]]. Task families are automatically node groups. Styling of a family node applies to all member nodes (tasks and sub-families), but precedence is determined by ordering in the suite definition. For example, if you style a family red and then one of its members green, cylc will plot a red family with one green member; but if you style one member green and then the family red, the red family styling will override the earlier green styling of the member.
[visualization] → [[node attributes]] → __NAME__ Replace __NAME__ with each node or node group for style attribute assignment.
This section defines all legal items and values for cylc site and user config files. See Site And User Config Files (Section 6) for file locations, intended usage, and how to generate the files using the cylc get-site-config command.
As for suite definitions, Jinja2 expressions can be embedded in site and user config files to generate the final result parsed by cylc. Use of Jinja2 in suite definitions is documented in Section 9.6.
A temporary directory is needed by a few cylc commands, and is cleaned automatically on exit. Leave unset for the default (usually $TMPDIR).
Number of process pool worker processes used to execute shell commands (job submission, event handlers, job poll and kill commands).
A rolling archive of suite state dumps is maintained under the suite run directory, and is used for restarts; this item determines the number of previous states retained. The most recent saved state file is called state. Sucessively older files have increasing integer values appended, starting from 1.
Commands that intervene in running suites can be made to ask for confirmation before acting. Some find this annoying and ineffective as a safety measure, however, so command prompts are disabled by default.
The suite run directory tree is created anew with every suite start (not restart) but output from the most recent previous runs can be retained in a rolling archive. Set length to 0 to keep no backups. This is incompatible with current Rose suite housekeeping (see Section 14 for more on Rose) so it is disabled by default, in which case new suite run files will overwrite existing ones in the same run directory tree. Rarely, this can result in incorrect polling results due to the presence of old task status files.
The number of old run directory trees to retain if run directory housekeeping is enabled.
Cylc can poll running jobs to catch problems that prevent task messages from being sent back to the suite, such as hard job kills, network outages, or unplanned task host shutdown. Routine polling is done only for the polling task communication method (below) unless suite-specific polling is configured in the suite definition. A list of interval values can be specified, with the last value used repeatedly until the task is finished - this allows more frequent polling near the beginning and end of the anticipated task run time. Multipliers can be used as shorthand as in the example below.
Cylc can also poll submitted jobs to catch problems that prevent the submitted job from executing at all, such as deletion from an external batch scheduler queue. Routine polling is done only for the polling task communication method (below) unless suite-specific polling is configured in the suite definition. A list of interval values can be specified as for execution polling (above) but a single value is probably sufficient for job submission polling.
This section contains configuration items that affect task-to-suite communications.
If a send fails, the messaging code will retry after a configured delay interval.
If successive sends fail, the messaging code will give up after a configured number of tries.
This is the same as the --pyro-timeout option in cylc commands. Without a timeout Pyro connections to unresponsive suites can hang indefinitely (suites suspended with Ctrl-Z for instance).
The suite event log, held under the suite run directory, is maintained as a rolling archive. Logs are rolled over (backed up and started anew) when they reach a configurable limit size.
If true, a new suite log will be started for a new suite run.
How many rolled logs to retain in the archive.
Suite event logs are rolled over when they reach this file size.
Documentation locations for the cylc doc command and gcylc Help menus.
File locations of documentation held locally on the cylc host server.
[documentation] → [files] → html index File location of the main cylc documentation index.
[documentation] → [files] → pdf user guide File location of the cylc User Guide, PDF version.
[documentation] → [files] → multi-page html user guide File location of the cylc User Guide, multi-page HTML version.
[documentation] → [files] → single-page html user guide File location of the cylc User Guide, single-page HTML version.
Online documentation URLs.
[documentation] → [urls] → internet homepage URL of the cylc internet homepage, with links to documentation for the latest official release.
[documentation] → [urls] → local index Local intranet URL of the main cylc documentation index.
PDF and HTML viewers can be launched by cylc to view the documentation.
Your preferred PDF viewer program.
Your preferred web browser.
Choose your favourite text editor for editing suite definitions.
The editor to be invoked by the cylc command line interface.
The editor to be invoked by the cylc GUI.
Pyro is the RPC layer used for network communication between cylc clients (suite-connecting commands and guis) servers (running suites). Each suite listens on a dedicated network port, binding on the first available starting at the configured base port.
The first port that cylc is allowed to use.
This determines the maximum number of suites that can run at once on the suite host.
Each suite stores its port number, by suite name, under this directory.
The [hosts] section configures some important host-specific settings for the suite host (‘localhost’) and remote task hosts. Note that remote task behaviour is determined by the site/user config on the suite host, not on the task host. Suites can specify task hosts that are not listed here, in which case local settings will be assumed, with the local home directory path, if present, replaced by $HOME in items that configure directory locations.
The default task host is the suite host, localhost, with default values as listed below. Use an explicit [hosts][[localhost]] section if you need to override the defaults. Localhost settings are then also used as defaults for other hosts, with the local home directory path replaced as described above. This applies to items omitted from an explicit host section, and to hosts that are not listed at all in the site and user config files. Explicit host sections are only needed if the automatically modified local defaults are not sufficient.
Host section headings can also be regular expressions to match multiple hostnames. Note that the general regular expression wildcard is ‘.⋆’ (zero or more of any character), not ‘⋆’. Hostname matching regular expressions are used as-is in the Python re.match() function. As such they match from the beginning of the hostname string (as specified in the suite definition) and they do not have to match through to the end of the string (use the string-end matching character ‘$’ in the expression to force this).
A hierachy of host match expressions from specific to general can be used because config items are processed in the order specified in the file.
[hosts] → HOST → run directory The top level of the directory tree that holds suite-specific output logs, state dump files, run database, etc.
[hosts] → HOST → work directory The top level for suite work and share directories.
[hosts] → HOST → task communication method The means by which task progress messages are reported back to the running suite. See above for default polling intervals for the poll method.
[hosts] → HOST → remote copy template A string for the command used to copy files to a remote host. This is not used on the suite host unless you run local tasks under another user account.
[hosts] → HOST → remote shell template A string template, containing \%s as a placeholder for the host name, for the command used to invoke commands on this host. This is not used on the suite host unless you run local tasks under another user account.
[hosts] → HOST → use login shell Whether to use a login shell or not for remote command invocation. By default cylc runs remote ssh commands using a login shell,
which will source /etc/profile and ~/.profile to set up the user environment. However, for security reasons some institutions do not allow unattended commands to start login shells, so you can turn off this behaviour to get,
which will use the default shell on the remote machine, sourcing ~/.bashrc (or ~/.cshrc) to set up the environment.
[hosts] → HOST → cylc executable The cylc executable on a remote host. Specify a full path if cylc is not in \$PATH when it is invoked via ssh on this host.
[hosts] → HOST → global initial scripting If specified, the value of this setting will be inserted to just before the initial scripting section of all job scripts that are to be submitted to the specified remote host.
[hosts] → HOST → copyable environment variables A list containing the names of the environment variables that can and/or need to be copied from the suite daemon to a job.
The suite host’s identity must be determined locally by cylc and passed to running tasks (via $CYLC_SUITE_HOST) so that task messages can target the right suite on the right host.
This item determines how cylc finds the identity of the suite host. For the default name method cylc asks the suite host for its host name. This should resolve on remote task hosts to the IP address of the suite host; if it doesn’t, adjust network settings or use one of the other methods. For the address method, cylc attempts to use a special external “target address” to determine the IP address of the suite host as seen by remote task hosts (in-source documentation in $CYLC_DIR/lib/cylc/suite_host.py explains how this works). And finally, as a last resort, you can choose the hardwired method and manually specify the host name or IP address of the suite host.
This item is required for the address self-identification method. If your suite host sees the internet, a common address such as google.com will do; otherwise choose a host visible on your intranet.
Use this item to explicitly set the name or IP address of the suite host if you have to use the hardwired self-identification method.
Utilities such as cylc gsummary need to scan hosts for running suites.
A list of hosts to scan for running suites.
This section defines all legal items and values for the gcylc user config file, which should be located in $HOME/.cylc/gcylc.rc. Current settings can be printed with the cylc get-gui-config command.
Set the suite view panel(s) displayed initially, when the GUI starts. This can be changed later using the tool bar.
List suite views, if any, that should be displayed initially in an ungrouped state. Namespace family grouping can be changed later using the tool bar.
Set the task state color theme, common to all views, to use initially. The color theme can be changed later using the tool bar. See gcylc.rc.eg and themes.rc in $CYLC_DIR/conf/gcylcrc/ for how to modify existing color themes or define your own. Use cylc get-gui-config to list your available themes.
Set the size of the task state dot icons displayed in the text and dot views.
If this is not turned off the default sort order for task names and families in the dot and text views will the order they appear in the suite definition. Clicking on the task name column in the treeview will toggle to alphanumeric sort, and a View menu item does the same for the dot view. If turned off, the default sort order is alphanumeric and definition order is not available at all.
This section may contain task state color theme definitions.
The name of the task state color-theme to be defined in this section.
You can inherit from another theme in order to avoid defining all states.
Set default icon attributes for all state icons in this theme.
For the attribute values, COLOR and FONTCOLOR are X color names or hex codes (e.g. #ff0000 for red); and STYLE can be “filled” or “unfilled”. See gcylc.rc.eg and themes.rc in $CYLC_DIR/conf/gcylcrc/ for examples.
Set icon attributes for all task states in THEME, or for a subset of them if you have used theme inheritance and/or defaults. Legal values of STATE are any of the cylc task proxy states: waiting, runahead, held, queued, ready, submitted, submit-failed, running, succeeded, failed, retrying, submit-retrying.
For the attribute values, COLOR and FONTCOLOR are X color names or hex codes (e.g. #ff0000 for red); and STYLE can be “filled” or “unfilled”. See gcylc.rc.eg and themes.rc in $CYLC_DIR/conf/gcylcrc/ for examples.
Cylc ("silk") is a suite engine and metascheduler that specializes in
cycling weather and climate forecasting suites and related processing
(but it can also be used for one-off workflows of non-cycling tasks).
For detailed documentation see the Cylc User Guide (cylc doc --help).
Version 6.1.1
The graphical user interface for cylc is "gcylc" (a.k.a. "cylc gui").
USAGE:
% cylc -v,--version # print cylc version
% cylc help,--help,-h,? # print this help page
% cylc help CATEGORY # print help by category
% cylc CATEGORY help # (ditto)
% cylc help [CATEGORY] COMMAND # print command help
% cylc [CATEGORY] COMMAND help,--help # (ditto)
% cylc [CATEGORY] COMMAND [options] SUITE [arguments]
% cylc [CATEGORY] COMMAND [options] SUITE TASK [arguments]
Commands and categories can both be abbreviated. Use of categories is
optional, but they organize help and disambiguate abbreviated commands:
% cylc control trigger SUITE TASK # trigger TASK in SUITE
% cylc trigger SUITE TASK # ditto
% cylc con trig SUITE TASK # ditto
% cylc c t SUITE TASK # ditto
CYLC SUITE NAMES AND YOUR REGISTRATION DATABASE
Suites are addressed by hierarchical names such as suite1, nwp.oper,
nwp.test.LAM2, etc. in a "name registration database" ($HOME/.cylc/REGDB)
that simply associates names with the suite definition locations. The
'--db=' command option can be used to view and copy suites from other
users, with access governed by normal filesystem permissions.
TASK IDENTIFICATION IN CYLC SUITES
Tasks are identified by NAME.CYCLE_POINT where POINT is either a
date-time or an integer.
Date-time cycle points are in an ISO 8601 date-time format, typically
CCYYMMDDThhmm followed by a time zone - e.g. 20101225T0600Z.
Integer cycle points (including those for one-off suites) are integers
- just '1' for one-off suites.
HOW TO DRILL DOWN TO COMMAND USAGE HELP:
% cylc help # list all available categories (this page)
% cylc help prep # list commands in category 'preparation'
% cylc help prep edit # command usage help for 'cylc [prep] edit'
Command CATEGORIES:
all ........... The complete command set.
db|database ... Suite name registration, copying, deletion, etc.
preparation ... Suite editing, validation, visualization, etc.
information ... Interrogate suite definitions and running suites.
discovery ..... Detect running suites.
control ....... Suite start up, monitoring, and control.
utility ....... Cycle arithmetic and templating, etc.
task .......... The task messaging interface.
hook .......... Suite and task event hook scripts.
admin ......... Cylc installation, testing, and example suites.
license|GPL ... Software licensing information (GPL v3.0).
CATEGORY: admin - Cylc installation, testing, and example suites.
HELP: cylc [admin] COMMAND help,--help
You can abbreviate admin and COMMAND.
The category admin may be omitted.
COMMANDS:
check-software .... Check required software is installed.
import-examples ... Import example suites your suite name database
test-battery ...... Run a battery of self-diagnosing test suites
test-db ........... Run an automated suite name database test
upgrade-db ........ Upgrade a pre-cylc-5.4 suite name database
upgrade-run-dir ... Upgrade a pre-cylc-6 suite run directory
CATEGORY: all - The complete command set.
HELP: cylc [all] COMMAND help,--help
You can abbreviate all and COMMAND.
The category all may be omitted.
COMMANDS:
5to6 ........................................ Improve the cylc 6 compatibility of a cylc 5 suite file
broadcast|bcast ............................. Change suite [runtime] settings on the fly
cat-log|log ................................. Print various suite and task log files
cat-state ................................... Print the state of tasks from the state dump
check-software .............................. Check required software is installed.
check-triggering ............................ A suite shutdown event hook for cylc testing
check-versions .............................. Compare cylc versions on task host accounts
conditions .................................. Print the GNU General Public License v3.0
copy|cp ..................................... Copy a suite or a group of suites
cycle-point|cyclepoint|datetime|cycletime ... Cycle point arithmetic and filename templating
depend ...................................... Add prerequisites to tasks in a running suite
diff|compare ................................ Compare two suite definitions and print differences
documentation|browse ........................ Display cylc documentation (User Guide etc.)
dump ........................................ Print the state of tasks in a running suite
edit ........................................ Edit suite definitions, optionally inlined
email-suite ................................. A suite event hook script that sends email alerts
email-task .................................. A task event hook script that sends email alerts
failed|task-failed .......................... (task messaging) Report task failed
get-directory ............................... Retrieve suite definition directory paths
get-gui-config .............................. Print gcylc configuration items
get-site-config|get-global-config ........... Print site/user configuration items
get-suite-config|get-config ................. Print suite configuration items
gpanel ...................................... Internal interface for GNOME 2 panel applet
graph ....................................... Plot suite dependency graphs and runtime hierarchies
gsummary .................................... Summary GUI for monitoring multiple suites
gui ......................................... (a.k.a. gcylc) cylc GUI for suite control etc.
hold ........................................ Hold (pause) suites or individual tasks
import-examples ............................. Import example suites your suite name database
insert ...................................... Insert tasks into a running suite
job-kill .................................... (Internal) Kill a job for a task
job-poll .................................... (Internal) Retrieve job status for a task
job-submit .................................. (Internal) Submit a job
jobscript ................................... Generate a task job script and print it to stdout
kill ........................................ Kill submitted or running tasks
list|ls ..................................... List suite tasks and family namespaces
message|task-message ........................ (task messaging) Report task messages
monitor ..................................... An in-terminal suite monitor (see also gcylc)
nudge ....................................... Cause the cylc task processing loop to be invoked
ping ........................................ Check that a suite is running
poll ........................................ Poll submitted or running tasks
print ....................................... Print registered suites
purge ....................................... Remove task trees from a running suite
random|rnd .................................. Generate a random integer within a given range
refresh ..................................... Report invalid registrations and update suite titles
register .................................... Register a suite for use
release|unhold .............................. Release (unpause) suites or individual tasks
reload ...................................... Reload the suite definition at run time
remove ...................................... Remove tasks from a running suite
reregister|rename ........................... Change the name of a suite
reset ....................................... Force one or more tasks to change state.
restart ..................................... Restart a suite from a previous state
run|start ................................... Start a suite at a given cycle point
scan ........................................ Scan a host for running suites
scp-transfer ................................ Scp-based file transfer for cylc suites
search|grep ................................. Search in suite definitions
set-runahead ................................ Change the runahead limit in a running suite.
set-verbosity ............................... Change a running suite's logging verbosity
show ........................................ Print task state (prerequisites and outputs etc.)
started|task-started ........................ (task messaging) Report task started
stop|shutdown ............................... Shut down running suites
submit|single ............................... Run a single task just as its parent suite would
succeeded|task-succeeded .................... (task messaging) Report task succeeded
suite-state ................................. Query the task states in a suite
test-battery ................................ Run a battery of self-diagnosing test suites
test-db ..................................... Run an automated suite name database test
trigger ..................................... Manually trigger or re-trigger a task
unregister .................................. Unregister and optionally delete suites
upgrade-db .................................. Upgrade a pre-cylc-5.4 suite name database
upgrade-run-dir ............................. Upgrade a pre-cylc-6 suite run directory
validate .................................... Parse and validate suite definitions
view ........................................ View suite definitions, inlined and Jinja2 processed
warranty .................................... Print the GPLv3 disclaimer of warranty
CATEGORY: control - Suite start up, monitoring, and control.
HELP: cylc [control] COMMAND help,--help
You can abbreviate control and COMMAND.
The category control may be omitted.
COMMANDS:
broadcast|bcast ... Change suite [runtime] settings on the fly
depend ............ Add prerequisites to tasks in a running suite
gui ............... (a.k.a. gcylc) cylc GUI for suite control etc.
hold .............. Hold (pause) suites or individual tasks
insert ............ Insert tasks into a running suite
kill .............. Kill submitted or running tasks
nudge ............. Cause the cylc task processing loop to be invoked
poll .............. Poll submitted or running tasks
purge ............. Remove task trees from a running suite
release|unhold .... Release (unpause) suites or individual tasks
reload ............ Reload the suite definition at run time
remove ............ Remove tasks from a running suite
reset ............. Force one or more tasks to change state.
restart ........... Restart a suite from a previous state
run|start ......... Start a suite at a given cycle point
set-runahead ...... Change the runahead limit in a running suite.
set-verbosity ..... Change a running suite's logging verbosity
stop|shutdown ..... Shut down running suites
trigger ........... Manually trigger or re-trigger a task
CATEGORY: db|database - Suite name registration, copying, deletion, etc.
Suite name registrations are held under $HOME/.cylc/REGDB.
HELP: cylc [db|database] COMMAND help,--help
You can abbreviate db|database and COMMAND.
The category db|database may be omitted.
COMMANDS:
copy|cp ............. Copy a suite or a group of suites
get-directory ....... Retrieve suite definition directory paths
print ............... Print registered suites
refresh ............. Report invalid registrations and update suite titles
register ............ Register a suite for use
reregister|rename ... Change the name of a suite
unregister .......... Unregister and optionally delete suites
CATEGORY: discovery - Detect running suites.
HELP: cylc [discovery] COMMAND help,--help
You can abbreviate discovery and COMMAND.
The category discovery may be omitted.
COMMANDS:
check-versions ... Compare cylc versions on task host accounts
ping ............. Check that a suite is running
scan ............. Scan a host for running suites
CATEGORY: hook - Suite and task event hook scripts.
HELP: cylc [hook] COMMAND help,--help
You can abbreviate hook and COMMAND.
The category hook may be omitted.
COMMANDS:
check-triggering ... A suite shutdown event hook for cylc testing
email-suite ........ A suite event hook script that sends email alerts
email-task ......... A task event hook script that sends email alerts
CATEGORY: information - Interrogate suite definitions and running suites.
HELP: cylc [information] COMMAND help,--help
You can abbreviate information and COMMAND.
The category information may be omitted.
COMMANDS:
cat-log|log ......................... Print various suite and task log files
cat-state ........................... Print the state of tasks from the state dump
documentation|browse ................ Display cylc documentation (User Guide etc.)
dump ................................ Print the state of tasks in a running suite
get-gui-config ...................... Print gcylc configuration items
get-site-config|get-global-config ... Print site/user configuration items
get-suite-config|get-config ......... Print suite configuration items
gpanel .............................. Internal interface for GNOME 2 panel applet
gsummary ............................ Summary GUI for monitoring multiple suites
gui|gcylc ........................... (a.k.a. gcylc) cylc GUI for suite control etc.
list|ls ............................. List suite tasks and family namespaces
monitor ............................. An in-terminal suite monitor (see also gcylc)
show ................................ Print task state (prerequisites and outputs etc.)
CATEGORY: license|GPL - Software licensing information (GPL v3.0).
HELP: cylc [license|GPL] COMMAND help,--help
You can abbreviate license|GPL and COMMAND.
The category license|GPL may be omitted.
COMMANDS:
conditions ... Print the GNU General Public License v3.0
warranty ..... Print the GPLv3 disclaimer of warranty
CATEGORY: preparation - Suite editing, validation, visualization, etc.
HELP: cylc [preparation] COMMAND help,--help
You can abbreviate preparation and COMMAND.
The category preparation may be omitted.
COMMANDS:
5to6 ........... Improve the cylc 6 compatibility of a cylc 5 suite file
diff|compare ... Compare two suite definitions and print differences
edit ........... Edit suite definitions, optionally inlined
graph .......... Plot suite dependency graphs and runtime hierarchies
jobscript ...... Generate a task job script and print it to stdout
list|ls ........ List suite tasks and family namespaces
search|grep .... Search in suite definitions
validate ....... Parse and validate suite definitions
view ........... View suite definitions, inlined and Jinja2 processed
CATEGORY: task - The task messaging interface.
HELP: cylc [task] COMMAND help,--help
You can abbreviate task and COMMAND.
The category task may be omitted.
COMMANDS:
failed|task-failed ......... (task messaging) Report task failed
job-kill ................... (Internal) Kill a job for a task
job-poll ................... (Internal) Retrieve job status for a task
job-submit ................. (Internal) Submit a job
message|task-message ....... (task messaging) Report task messages
started|task-started ....... (task messaging) Report task started
submit|single .............. Run a single task just as its parent suite would
succeeded|task-succeeded ... (task messaging) Report task succeeded
CATEGORY: utility - Cycle arithmetic and templating, etc.
HELP: cylc [utility] COMMAND help,--help
You can abbreviate utility and COMMAND.
The category utility may be omitted.
COMMANDS:
cycle-point|cyclepoint|datetime|cycletime ... Cycle point arithmetic and filename templating
random|rnd .................................. Generate a random integer within a given range
scp-transfer ................................ Scp-based file transfer for cylc suites
suite-state ................................. Query the task states in a suite
USAGE: cylc [prep] 5to6 FILE
Suggest changes to a cylc 5 suite file to make it more cylc 6 compatible.
This may be a suite.rc file, an include file, or a suite.rc.processed file.
By default, print the changed file to stdout. Lines that have been changed
are marked with '# UPGRADE'. These marker comments are purely for your own
information and should not be included in any changes you make. In
particular, they may break continuation lines.
Lines with '# UPGRADE CHANGE' have been altered.
Lines with '# UPGRADE ... INFO' indicate that manual change is needed.
You may want to run 'cylc validate' on your suite to see the automatic
start-up/mixed async replacement R1⋆ section(s). The validity of these can
be highly dependent on the initial cycle point choice (e.g. whether it is
T00 or T12).
This command works best for hour-based cycling - it will always convert
e.g. 'foo[T-6]' to 'foo[-PT6H]', even where this is in a monthly or yearly
cycling section graph.
This command is an aid, and is not an auto-upgrader or a substitute for
reading the documentation. The suggested changes must be understood and
checked by hand.
Example usage:
# Print out a file path (FILE) with suggested changes to stdout.
cylc 5to6 FILE
# Replace the file with the suggested changes file.
cylc 5to6 FILE > FILE
# Save a copy of the changed file.
cylc 5to6 FILE > FILE.5to6
# Show the diff of the changed file vs the original file.
diff - <(cylc 5to6 FILE) <FILE
Options:
-h, --help Print this help message and exit.
Usage: cylc [control] broadcast|bcast [OPTIONS] REG
Override [runtime] config in targeted namespaces in a running suite.
Uses for broadcast include making temporary changes to task behaviour,
and task-to-downstream-task communication via environment variables.
A broadcast can target any [runtime] namespace for all cycles or for a
specific cycle. If a task is affected by specific-cycle and all-cycle
broadcasts at once, the specific takes precedence. If a task is affected
by broadcasts to multiple ancestor namespaces, the result is determined
by normal [runtime] inheritance. In other words, it follows this order:
all:root -> all:FAM -> all:task -> tag:root -> tag:FAM -> tag:task
Broadcasts persist, even across suite restarts, until they expire when
their target cycle point is older than the oldest current in the suite,
or until they are explicitly cancelled with this command. All-cycle
broadcasts do not expire.
For each task the final effect of all broadcasts to all namespaces is
computed on the fly just prior to job submission. The --cancel and
--clear options simply cancel (remove) active broadcasts, they do not
act directly on the final task-level result. Consequently, for example,
you cannot broadcast to "all cycles except Tn" with an all-cycle
broadcast followed by a cancel to Tn (there is no direct broadcast to Tn
to cancel); and you cannot broadcast to "all members of FAMILY except
member_n" with a general broadcast to FAMILY followed by a cancel to
member_n (there is no direct broadcast to member_n to cancel).
To broadcast a variable to all tasks (quote items with internal spaces):
% cylc broadcast -s "[environment]VERSE = the quick brown fox" REG
To cancel the same broadcast:
% cylc broadcast --cancel "[environment]VERSE" REG
Use -d/--display to see active broadcasts. Multiple set or cancel
options can be used on the same command line. Broadcast cannot change
[runtime] inheritance.
See also 'cylc reload' - reload a modified suite definition at run time.
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
-t CYCLE_POINT, --tag=CYCLE_POINT
(Deprecated). Target cycle point. More than one can be
added. Defaults to 'all-cycle-points' with --set and
--cancel, and nothing with --clear.
-p CYCLE_POINT, --point=CYCLE_POINT
Target cycle point. More than one can be added.
Defaults to 'all-cycle-points' with --set and
--cancel, and nothing with --clear.
-n NAME, --namespace=NAME
Target namespace. Defaults to 'root' with --set and
--cancel, and nothing with --clear.
-s [SEC]ITEM=VALUE, --set=[SEC]ITEM=VALUE
A [runtime] config item and value to broadcast.
-c [SEC]ITEM, --cancel=[SEC]ITEM
An item-specific broadcast to cancel.
-C, --clear Cancel all broadcasts, or with -p/--point,
-n/--namespace, cancel all broadcasts to targeted
namespaces and/or cycle points. Use '-C -p all-cycle-
points' to cancel all all-cycle broadcasts without
canceling all specific-cycle broadcasts.
-e CYCLE, --expire=CYCLE
Cancel any broadcasts that target cycle times earlier
than, but not inclusive of, CYCLE.
-d, --display Display active broadcasts.
-k TASKID, --display-task=TASKID
Print active broadcasts for a given task
(NAME.CYCLE_POINT).
-b, --box Use unicode box characters with -d, -k.
-r, --raw With -d/--display or -k/--display-task, write out the
broadcast config structure in raw Python form.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands.
The default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that
it is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [info] cat-log|log [OPTIONS] REG [TASK-ID]
Print various log files for suites and tasks that are currently running,
or have previously finished.
Arguments:
REG Suite name
[TASK-ID] Print the stdout or stderr log
of the identified task
Options:
-h, --help show this help message and exit
-l, --location Just print the location of the requested log file.
-r INT, --rotation=INT
Rotation number (to view older, rotated suite logs)
-o, --stdout Print suite or task stdout logs (for suites, the
default is to print the event log; for tasks, the
default is to print the job script).
-e, --stderr Print suite or task stderr logs (see --stdout for
defaults).
-t INT, --try-number=INT
Task try number (default 1).
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
Usage: cylc [info] cat-state [OPTIONS] REG
Print the suite state dump file directly to stdout.
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
-d, --dump Use the same display format as the 'cylc dump' command.
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
USAGE: cylc [hook] check-triggering ARGS
This is a cylc shutdown event handler that compares the newly generated
suite log with a previously generated reference log "reference.log"
stored in the suite definition directory. Currently it just compares
runtime triggering information, disregarding event order and timing, and
fails the suite if there is any difference. This should be sufficient to
verify correct scheduling of any suite that is not affected by different
run-to-run conditional triggering.
1) run your suite with "cylc run --generate-reference-log" to generate
the reference log with resolved triggering information. Check manually
that the reference run was correct.
2) run reference tests with "cylc run --reference-test" - this
automatically sets the shutdown event handler along with a suite timeout
and "abort if shutdown handler fails", "abort on timeout", and "abort if
any task fails".
Reference tests can use any run mode:
⋆ simulation mode - tests that scheduling is equivalent to the reference
⋆ dummy mode - also tests that task hosting, job submission, job script
evaluation, and cylc messaging are not broken.
⋆ live mode - tests everything (but takes longer with real tasks!)
If any task fails, or if cylc itself fails, or if triggering is not
equivalent to the reference run, the test will abort with non-zero exit
status - so reference tests can be used as automated tests to check
that changes to cylc have not broken your suites.
Usage: cylc [discovery] check-versions [OPTIONS] SUITE
Check the version of cylc invoked on each of SUITE's task host accounts
when CYLC_VERSION is set to 6.1.1 (i.e. ⋆this⋆ version).
Different versions are reported but are not considered an error unless
the -e|--error option is specified, because different cylc versions are
not necessarily (nor usually) incompatible.
It is recommended that cylc versions be installed in parallel and access
configured via the cylc version wrapper as described in the cylc INSTALL
file and User Guide. Users then get the latest installed version by
default, or (like tasks) a particular version if $CYLC_VERSION is defined.
Remote cylc versions are interrogated like this:
ssh user@host \
"bash --login -c 'CYLC_VERSION=6.1.1 cylc -v'"
A login shell is used because task job scripts currently source login
scripts explicitly at start-up to configure access to cylc.
Arguments:
SUITE Suite name or path
Options:
-h, --help show this help message and exit
-e, --error Exit with error status if 6.1.1 is not available on
all remote accounts.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [db] copy|cp [OPTIONS] REG REG2 TOPDIR
Copy suite or group REG to TOPDIR, and register the copy as REG2.
Consider the following three suites:
% cylc db print '^foo' # printed in flat form
foo.bag | "Test Suite Zero" | /home/bob/zero
foo.bar.qux | "Test Suite Two" | /home/bob/two
foo.bar.baz | "Test Suite One" | /home/bob/one
% cylc db print -t '^foo' # printed in tree from
foo
|-bag "Test Suite Zero" | /home/bob/zero
‘-bar
|-baz "Test Suite One" | /home/bob/one
‘-qux "Test Suite Two" | /home/bob/two
These suites are stored in a flat directory structure under /home/bob,
but they are organised in the suite database as a group 'foo' that
contains the suite 'foo.bag' and a group 'foo.bar', which in turn
contains the suites 'foo.bar.baz' and 'foo.bar.qux'.
When you copy suites with this command, the target registration names
are determined by TARGET and the name structure underneath SOURCE, and
the suite definition directories are copied into a directory tree under
TOPDIR whose structure reflects the target registration names. If this
is not what you want, you can copy suite definition directories manually
and then register the copied directories manually with 'cylc register'.
EXAMPLES (using the three suites above):
% cylc db copy foo.bar.baz red /home/bob # suite to suite
Copying suite definition for red
% cylc db print "^red"
red | "Test Suite One" | /home/bob/red
% cylc copy foo.bar.baz blue.green /home/bob # suite to group
Copying suite definition for blue.green
% cylc db pr "^blue"
blue.green | "Test Suite One" | /home/bob/blue/green
% cylc copy foo.bar orange /home/bob # group to group
Copying suite definition for orange.qux
Copying suite definition for orange.baz
% cylc db pr "^orange"
orange.qux | "Test Suite Two" | /home/bob/orange/qux
orange.baz | "Test Suite One" | /home/bob/orange/baz
Arguments:
REG Source suite name
REG2 Target suite name
TOPDIR Top level target directory.
Options:
-h, --help show this help message and exit
--db-from=PATH Copy suites from another DB (defaults to --db).
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location, defaults
to $HOME/.cylc/REGDB.
Usage: cylc [util] cycle-point [OPTIONS] [POINT]
Cycle point date-time offset computation, and filename templating.
Filename templating replaces elements of a template string with corresponding
elements of the current or given cycle point.
The template string format changed at cylc-6, but is backward compatible.
If a 10-digit pre cylc-6 cycle time is detected, the template string
elements are 'YYYY' (year), 'MM' (month), 'DD' (day), and 'HH' (hour):
% cylc cyclepoint 2010080800 --template foo-YYYY-MM-DD-HH.nc
foo-2010-08-08-00.nc
Otherwise (cylc-6+) use ISO 8601 or posix date-time format elements:
% cylc cyclepoint 2010080T800 --template foo-CCYY-MM-DD-Thh.nc
foo-2010-08-08-T00.nc
% cylc cyclepoint 2010080T800 --template foo-%Y-%m-%d-Thh.nc
foo-2010-08-08-T00.nc
Other examples:
1) print offset from an explicit cycle point:
% cylc [util] cycle-point --offset-hours=6 20100823T1800Z
20100824T0000Z
2) print offset from $CYLC_TASK_CYCLE_POINT (as in suite tasks):
% export CYLC_TASK_CYCLE_POINT=20100823T1800Z
% cylc cycle-point --offset-hours=-6
20100823T1200Z
3) cycle point filename templating, explicit template:
% export CYLC_TASK_CYCLE_POINT=2010-08
% cylc cycle-point --offset-years=2 --template=foo-CCYY-MM.nc
foo-2012-08.nc
4) cycle point filename templating, template in a variable:
% export CYLC_TASK_CYCLE_POINT=2010-08
% export MYTEMPLATE=foo-CCYY-MM.nc
% cylc cycle-point --offset-years=2 --template=MYTEMPLATE
foo-2012-08.nc
Arguments:
[CYCLE] ISO 8601 date-time, e.g. 20140201T0000Z, default
$CYLC_TASK_CYCLE_POINT
Options:
-h, --help show this help message and exit
--offset-hours=HOURS Add N hours to CYCLE (may be negative)
--offset-days=DAYS Add N days to CYCLE (N may be negative)
--offset-months=MONTHS
Add N months to CYCLE (N may be negative)
--offset-years=YEARS Add N years to CYCLE (N may be negative)
--offset=ISO_OFFSET Add an ISO 8601-based interval representation to CYCLE
--template=TEMPLATE Filename template string or variable
--time-zone=TEMPLATE Control the formatting of the result's timezone e.g.
(Z, +13:00, -hh
--num-expanded-year-digits=NUMBER
Specify a number of expanded year digits to print in
the result
--print-year Print only CCYY of result
--print-month Print only MM of result
--print-day Print only DD of result
--print-hour Print only HH of result
Usage: cylc [control] depend [OPTIONS] REG TASK DEP
Add new dependencies on the fly to tasks in running suite REG. If DEP
is a task ID the target TASK will depend on that task finishing,
otherwise DEP can be an explicit quoted message such as
"Data files uploaded for 2011080806"
(presumably there will be another task in the suite, or you will insert
one, that reports that message as an output).
Prerequisites added on the fly are not propagated to the successors
of TASK, and they will not persist in TASK across a suite restart.
Arguments:
REG Suite name
TASK Target task
DEP New dependency
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [prep] diff|compare [OPTIONS] SUITE1 SUITE2
Compare two suite definitions and display any differences.
Differencing is done after parsing the suite.rc files so it takes
account of default values that are not explicitly defined, it disregards
the order of configuration items, and it sees any include-file content
after inlining has occurred.
Note that seemingly identical suites normally differ due to inherited
default configuration values (e.g. the default job submission log
directory.
Files in the suite bin directory and other sub-directories of the
suite definition directory are not currently differenced.
Arguments:
SUITE1 Suite name or path
SUITE2 Suite name or path
Options:
-h, --help show this help message and exit
-n, --nested print suite.rc section headings in nested form.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [info] documentation|browse [OPTIONS]
By default this command opens the cylc documentation index in your
browser in file:// mode. Alternatively it can open the PDF Cylc User
Guide directly, or browse the cylc internet homepage, or - if your site
has a web server with access to the cylc documentation - an intranet
documentation URL. The browser and PDF reader to use, and the intranet
URL, is determined by cylc site/user configuration - for details see
$ cylc site-config --help
Options:
-h, --help show this help message and exit
-p, --pdf Open the PDF User Guide directly
-w, --internet Browse the cylc internet homepage
Usage: cylc [info] dump [OPTIONS] REG
Print state information (e.g. the state of each task) from a running
suite. For small suites 'watch cylc [info] dump SUITE' is an effective
non-GUI real time monitor (but see also 'cylc monitor').
For more information about a specific task, such as the current state of
its prerequisites and outputs, see 'cylc [info] show'.
Examples:
Display the state of all running tasks, sorted by cycle point:
% cylc [info] dump --tasks --sort SUITE | grep running
Display the state of all tasks in a particular cycle point:
% cylc [info] dump -t SUITE | grep 2010082406
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
-g, --global Global information only.
-t, --tasks Task states only.
-s, --sort Task states only; sort by cycle point instead of name.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
Usage: cylc [prep] edit [OPTIONS] SUITE
Edit suite definitions without having to move to their directory
locations, and with optional reversible inlining of include-files. Note
that Jinja2 suites can only be edited in raw form but the processed
version can be viewed with 'cylc [prep] view -p'.
1/cylc [prep] edit SUITE
Change to the suite definition directory and edit the suite.rc file.
2/ cylc [prep] edit -i,--inline SUITE
Edit the suite with include-files inlined between special markers. The
original suite.rc file is temporarily replaced so that the inlined
version is "live" during editing (i.e. you can run suites during
editing and cylc will pick up changes to the suite definition). The
inlined file is then split into its constituent include-files
again when you exit the editor. Include-files can be nested or
multiply-included; in the latter case only the first inclusion is
inlined (this prevents conflicting changes made to the same file).
3/ cylc [prep] edit --cleanup SUITE
Remove backup files left by previous INLINED edit sessions.
INLINED EDITING SAFETY: The suite.rc file and its include-files are
automatically backed up prior to an inlined editing session. If the
editor dies mid-session just invoke 'cylc edit -i' again to recover from
the last saved inlined file. On exiting the editor, if any of the
original include-files are found to have changed due to external
intervention during editing you will be warned and the affected files
will be written to new backups instead of overwriting the originals.
Finally, the inlined suite.rc file is also backed up on exiting
the editor, to allow recovery in case of accidental corruption of the
include-file boundary markers in the inlined file.
The edit process is spawned in the foreground as follows:
% <editor> suite.rc
Where <editor> is defined in the cylc site/user config files.
See also 'cylc [prep] view'.
Arguments:
SUITE Suite name or path
Options:
-h, --help show this help message and exit
-i, --inline Edit with include-files inlined as described above.
--cleanup Remove backup files left by previous inlined edit sessions.
-g, --gui Force use of the configured GUI editor.
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
USAGE: cylc [hook] email-suite EVENT SUITE MESSAGE
This is a simple suite event hook script that sends an email.
The command line arguments are supplied automatically by cylc.
For example, to get an email alert when a suite shuts down:
# SUITE.RC
[cylc]
[[environment]]
MAIL_ADDRESS = foo@bar.baz.waz
[[event hooks]]
shutdown handler = cylc email-suite
See the Suite.rc Reference (Cylc User Guide) for more information
on suite and task event hooks and event handler scripts.
USAGE: cylc [hook] email-task EVENT SUITE TASKID MESSAGE
This is a simple task event hook handler script that sends an email.
The command line arguments are supplied automatically by cylc.
For example, to get an email alert whenever any task fails:
# SUITE.RC
[cylc]
[[environment]]
MAIL_ADDRESS = foo@bar.baz.waz
[runtime]
[[root]]
[[[event hooks]]]
failed handler = cylc email-task
See the Suite.rc Reference (Cylc User Guide) for more information
on suite and task event hooks and event handler scripts.
Usage: cylc [task] failed [OPTIONS] [REASON]
This command is part of the cylc task messaging interface, used by
running tasks to communicate progress to their parent suite.
The failed command reports to successfully complete task execution.
It is automatically called by task job scripts, except in the case of
"detaching tasks" which must do their own completion messaging.
Suite and task identity are determined from the task execution
environment supplied by the suite (or by the single task 'submit'
command, in which case case the message is just printed to stdout).
See also:
cylc [task] message
cylc [task] started
cylc [task] succeeded
Arguments:
REASON - message explaining why the task failed.
Options:
-h, --help show this help message and exit
-v, --verbose Verbose output mode.
Usage: cylc [db] get-directory REG
Retrieve and print the directory location of suite REG.
Here's an easy way to move to a suite directory:
$ cd $(cylc get-dir REG).
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [admin] get-gui-config [OPTIONS]
Print gcylc configuration settings.
By default all settings are printed. For specific sections or items
use -i/--item and wrap parent sections in square brackets:
cylc get-gui-config --item '[themes][default]succeeded'
Multiple items can be specified at once.
Options:
-h, --help show this help message and exit
-v, --verbose Print extra information.
--debug Show exception tracebacks.
-i [SEC...]ITEM, --item=[SEC...]ITEM
Item or section to print (multiple use allowed).
--sparse Only print items explicitly set in the config files.
-p, --python Print native Python format.
Usage: cylc [admin] get-site-config [OPTIONS]
Print cylc site/user configuration settings.
By default all settings are printed. For specific sections or items
use -i/--item and wrap parent sections in square brackets:
cylc get-site-config --item '[editors]terminal'
Multiple items can be specified at once.
Options:
-h, --help show this help message and exit
-i [SEC...]ITEM, --item=[SEC...]ITEM
Item or section to print (multiple use allowed).
--sparse Only print items explicitly set in the config files.
-p, --python Print native Python format.
--print-run-dir Print the configured cylc run directory.
-v, --verbose Print extra information.
--debug Show exception tracebacks.
Usage: cylc [info] get-suite-config [OPTIONS] SUITE
Print parsed suite configuration items, after runtime inheritance.
By default all settings are printed. For specific sections or items
use -i/--item and wrap sections in square brackets, e.g.:
cylc get-suite-config --item '[scheduling]initial cycle point'
Multiple items can be retrieved at once.
By default, unset values are printed as an empty string, or (for
historical reasons) as "None" with -o/--one-line. These defaults
can be changed with the -n/--null-value option.
Example:
|# SUITE.RC
|[runtime]
| [[modelX]]
| [[[environment]]]
| FOO = foo
| BAR = bar
$ cylc get-suite-config --item=[runtime][modelX][environment]FOO SUITE
foo
$ cylc get-suite-config --item=[runtime][modelX][environment] SUITE
FOO = foo
BAR = bar
$ cylc get-suite-config --item=[runtime][modelX] SUITE
...
[[[environment]]]
FOO = foo
BAR = bar
...
Arguments:
SUITE Suite name or path
Options:
-h, --help show this help message and exit
-i [SEC...]ITEM, --item=[SEC...]ITEM
Item or section to print (multiple use allowed).
-r, --sparse Only print items explicitly set in the config files.
-p, --python Print native Python format.
-a, --all-tasks For [runtime] items (e.g. --item='command scripting')
report values for all tasks prefixed by task name.
-n STRING, --null-value=STRING
The string to print for unset values (default
nothing).
-m, --mark-up Prefix each line with '!cylc!'.
-o, --one-line Print multiple single-value items at once.
-t, --tasks Print the suite task list [DEPRECATED: use 'cylc list
SUITE'].
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc gpanel [OPTIONS]
This is a cylc summary panel applet for monitoring running suites on a set of
hosts in GNOME 2.
To install this applet, run "cylc gpanel --install"
and follow the instructions that it gives you.
This applet can be tested using the --test option.
To customize themes, copy $CYLC_DIR/conf/gcylcrc/gcylc.rc.eg to
$HOME/.cylc/gcylc.rc and follow the instructions in the file.
To configure default suite hosts, edit the
[suite host scanning]hosts entry in your global.rc file.
Options:
-h, --help show this help message and exit
--compact Switch on compact mode at runtime.
--install Install the panel applet.
--test Run in a standalone window.
Usage: 1/ cylc [prep] graph [OPTIONS] SUITE [START[STOP]]
Plot the suite.rc dependency graph for SUITE.
2/ cylc [prep] graph [OPTIONS] -f,--file FILE
Plot the specified dot-language graph file.
Plot suite dependency graphs in an interactive graph viewer.
If START is given it overrides "[visualization] initial cycle point" to
determine the start point of the graph, which defaults to the suite initial
cycle point. If STOP is given it overrides "[visualization] final cycle point"
to determine the end point of the graph, which defaults to the graph start
point plus "[visualization] number of cycle points" (which defaults to 3).
The graph start and end points are adjusted up and down to the suite initial
and final cycle points, respectively, if necessary.
The "Save" button generates an image of the current view, of format (e.g. png,
svg, jpg, eps) determined by the filename extension. If the chosen format is
not available a dialog box will show those that are available.
If the optional output filename is specified, the viewer will not open and a
graph will be written directly to the file.
GRAPH VIEWER CONTROLS:
⋆ Center on a node: left-click.
⋆ Pan view: left-drag.
⋆ Zoom: +/- buttons, mouse-wheel, or ctrl-left-drag.
⋆ Box zoom: shift-left-drag.
⋆ "Best Fit" and "Normal Size" buttons.
⋆ Left-to-right graphing mode toggle button.
⋆ "Ignore suicide triggers" button.
⋆ "Ignore cold-start tasks" button.
⋆ "Save" button: save an image of the view.
Family (namespace) grouping controls:
Toolbar:
⋆ "group" - group all families up to root.
⋆ "ungroup" - recursively ungroup all families.
Right-click menu:
⋆ "group" - close this node's parent family.
⋆ "ungroup" - open this family node.
⋆ "recursive ungroup" - ungroup all families below this node.
Arguments:
[SUITE] Suite name or path
[START] Initial cycle point to plot (default=2999010100)
[STOP] Final cycle point to plot (default=2999010123)
Options:
-h, --help show this help message and exit
-n, --namespaces Plot the suite namespace inheritance hierarchy (task
run time properties).
-f FILE, --file=FILE View a specific dot-language graphfile.
-O FILE, --output-file=FILE
Output to a specific file, with a format given by
--output-format or extrapolated from the extension.
--output-format=FORMAT
Specify a format for writing out the graph to
--output-file e.g. png, svg, jpg, eps, dot.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc gsummary [OPTIONS]
cylc gsummary [OPTIONS]
This is the cylc summary gui for monitoring running suites on a set of
hosts.
To customize themes copy $CYLC_DIR/conf/gcylcrc/gcylc.rc.eg to
$HOME/.cylc/gcylc.rc and follow the instructions in the file.
Options:
-h, --help show this help message and exit
--user=USER User account name (defaults to $USER).
--host=HOST Host names to monitor (override site default).
--poll-interval=SECONDS
Polling interval (time between updates) in seconds
Usage: cylc gui [OPTIONS] [REG]
gcylc [OPTIONS] [REG]
This is the cylc Graphical User Interface.
Local suites can be opened and switched between from within gcylc. To
connect to running remote suites (whose passphrase you have installed)
you must currently use --host and/or --user on the gcylc command line.
Available task state color themes are shown under the View menu. To
customize themes copy $CYLC_DIR/conf/gcylcrc/gcylc.rc.eg to
$HOME/.cylc/gcylc.rc and follow the instructions in the file.
To see current configuration settings use "cylc get-gui-config".
Arguments:
[REG] Suite name
Options:
-h, --help show this help message and exit
-r, --restricted Restrict display to 'active' task states: submitted,
submit-failed, submit-retrying, running, failed,
retrying; and disable the graph view. This may be
needed for very large suites. The state summary icons
in the status bar still represent all task proxies.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands.
The default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [control] hold [OPTIONS] REG [MATCH] [POINT]
Hold one or more waiting tasks, or a whole suite. Held tasks do not
submit even if they are ready to run.
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
See also 'cylc [control] release'.
Arguments:
REG Suite name
[MATCH] Task or family name matching regular expression
[POINT] Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [control] insert [OPTIONS] REG MATCH CYCLE_POINT [STOP_POINT]
Insert task proxies into a running suite. Uses of insertion include:
1) insert a task that was excluded by the suite definition at start-up.
2) reinstate a task that was previously removed from a running suite.
3) re-run an old task that cannot be retriggered because its task proxy
is no longer live in the a suite.
Be aware that inserted cycling tasks keep on cycling as normal, even if
another instance of the same task exists at a later cycle (instances of
the same task at different cycles can coexist, but a newly spawned task
will not be added to the pool if it catches up to another task with the
same ID).
See also 'cylc submit', for running tasks without the scheduler.
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
Arguments:
REG Suite name
MATCH Task or family name matching regular expression
CYCLE_POINT Cycle point (e.g. date-time or integer)
[STOP_POINT] Optional stop cycle point for inserted task.
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [control] job-kill ST-FILE
(This command is for internal use. Users should use "cylc kill".) Read the job
status file to obtain the name of the batch system and the job ID in the
system. Invoke the relevant batch system command to ask the batch system to
remove and terminate the job.
Arguments:
ST-FILE the task status file
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [control] job-poll ST-FILE
(This command is for internal use. Users should use "cylc poll".) Poll a
submitted or running job by inspecting its status file and possibly querying
the batch system to see if the job is still alive or not.
Arguments:
ST-FILE the task status file
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [task] job-submit [--remote-mode] JOB-FILE-PATH
(This command is for internal use. Users should use "cylc submit".)
Submit a job file.
Arguments:
JOB-FILE-PATH the path of the job file
Options:
-h, --help show this help message and exit
--remote-mode Is this being run on a remote job host?
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
USAGE: cylc [prep] jobscript [OPTIONS] REG TASK
Generate a task job script and print it to stdout.
Here's how to capture the script in the vim editor:
% cylc jobscript REG TASK | vim -
Emacs unfortunately cannot read from stdin:
% cylc jobscript REG TASK > tmp.sh; emacs tmp.sh
This command wraps 'cylc [control] submit --dry-run'.
Other options (e.g. for suite host and owner) are passed
through to the submit command.
Options:
-h,--help - print this usage message.
(see also 'cylc submit --help')
Arguments:
REG - Registered suite name.
TASK - Task ID (NAME.CYCLE_POINT)
Usage: cylc [control] kill [OPTIONS] REG MATCH [POINT]
Kill a 'submitted' or 'running' task and update the suite state accordingly.
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
Arguments:
REG Suite name
MATCH Task or family name matching regular expression
[POINT] Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [info|prep] list|ls [OPTIONS] SUITE
Print runtime namespace names (tasks and families), the first-parent
inheritance graph, or actual tasks for a given cycle range.
The first-parent inheritance graph determines the primary task family
groupings that are collapsible in gcylc suite views and the graph
viewer tool. To visualize the full multiple inheritance hierarchy use:
'cylc graph -n'.
Arguments:
SUITE Suite name or path
Options:
-h, --help show this help message and exit
-a, --all-tasks Print all tasks, not just those used in the graph.
-n, --all-namespaces Print all runtime namespaces, not just tasks.
-m, --mro Print the linear "method resolution order" for each
namespace (the multiple-inheritance precedence order
as determined by the C3 linearization algorithm).
-t, --tree Print the first-parent inheritance hierarchy in tree
form.
-b, --box With -t/--tree, using unicode box characters. Your
terminal must be able to display unicode characters.
-w, --with-titles Print namespaces titles too.
-p START[,STOP], --points=START[,STOP]
Print actual task IDs from the START [through STOP]
cycle points.
-c START[,STOP], --cycles=START[,STOP]
(deprecated: use -p/--points).
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [task] message [OPTIONS] MESSAGE
This command is part of the cylc task messaging interface, used by
running tasks to communicate progress to their parent suite.
The message command can be used to report "message outputs" completed.
Other messages received by the suite daemon will just be logged.
Suite and task identity are determined from the task execution
environment supplied by the suite (or by the single task 'submit'
command, in which case case the message is just printed to stdout).
See also:
cylc [task] started
cylc [task] succeeded
cylc [task] failed
Options:
-h, --help show this help message and exit
-p PRIORITY message priority: NORMAL, WARNING, or CRITICAL;
default NORMAL.
--next-restart-completed
Report next restart file(s) completed
--all-restart-outputs-completed
Report all restart outputs completed at once.
--all-outputs-completed
Report all internal outputs completed at once.
-v, --verbose Verbose output mode.
Usage: cylc [info] monitor [OPTIONS] REG
A terminal-based suite monitor.
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
-a, --align Align task names. Only useful for small suites.
-r, --restricted Restrict display to 'active' task states: submitted,
submit-failed, submit-retrying, running, failed,
retrying. This may be needed for very large suites. The
state summary line still represents all task proxies.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
Usage: cylc [control] nudge [OPTIONS] REG
Cause the cylc task processing loop to be invoked in a running suite.
This happens automatically when the state of any task changes such that
task processing (dependency negotation etc.) is required, or if a
clock-triggered task is ready to run.
The main reason to use this command is to update the "estimated time till
completion" intervals shown in the tree-view suite control GUI, during
periods when nothing else is happening.
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [discovery] ping [OPTIONS] REG [TASK]
If suite REG (or task TASK in it) is running, exit (silently, unless
-v,--verbose is specified); else print an error message and exit with
error status. For tasks, success means the task proxy is currently in
the 'running' state.
Arguments:
REG Suite name
[TASK] Task NAME.CYCLE_POINT
Options:
-h, --help show this help message and exit
--print-ports Print the port range from the cylc site config file.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [control] poll [OPTIONS] REG MATCH POINT
Poll a 'submitted' or 'running' task to verify its status. If a job was
killed by external means this will update the suite accordingly.
Note that automatic job polling can used to track task status on task
hosts that do not allow any communication by RPC (pyro) or ssh back to
the suite host - see site/user config file documentation.
Polling is also done automatically on restarting a suite, for any tasks
that were recorded as submitted or running when the suite went down.
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
Arguments:
REG Suite name
MATCH Task or family name matching regular expression
POINT Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [db] print [OPTIONS] [REGEX]
Print suite database registrations.
Note on result filtering:
(a) The filter patterns are Regular Expressions, not shell globs, so
the general wildcard is '.⋆' (match zero or more of anything), NOT '⋆'.
(b) For printing purposes there is an implicit wildcard at the end of
each pattern ('foo' is the same as 'foo.⋆'); use the string end marker
to prevent this ('foo$' matches only literal 'foo').
Arguments:
[REGEX] Suite name regular expression pattern
Options:
-h, --help show this help message and exit
-t, --tree Print registrations in nested tree form.
-b, --box Use unicode box drawing characters in tree views.
-a, --align Align columns.
-x don't print suite definition directory paths.
-y Don't print suite titles.
--fail Fail (exit 1) if no matching suites are found.
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [control] purge [OPTIONS] REG TASK STOP_POINT
Remove an entire tree of dependent tasks, over multiple cycles, from a running
suite. The top task will be forced to spawn and will then be removed, then so
will every task that depends on it, and every task that depends on those, and
so on until the given stop cycle point.
THIS COMMAND IS DANGEROUS.
UNDERSTANDING HOW PURGE WORKS: cylc identifies tasks that depend on the top
task, and then on its downstream dependents, and then on theirs, etc., by
simulating what would happen if the top task were to trigger: it artificially
sets the top task to the "succeeded" state then negotatiates dependencies and
artificially sets any tasks whose prerequisites get satisfied to "succeeded";
then it negotiates dependencies again, and so on until the stop cycle is
reached or nothing new triggers. Finally it marks "virtually triggered" tasks
for removal. Consequently:
⋆ Dependent tasks will only be identified as such, and purged, if they have
already spawned into the top cycle - so let them catch up first.
⋆ You can't purge a tree of tasks that has already triggered, because the
algorithm relies on detecting new triggering.
Note that suite runahead must be sufficient enough to bridge the purge gap.
Arguments:
REG Suite name
TASK Task (NAME.CYCLE_POINT) to start purge
STOP_POINT Cycle point (inclusive!) to stop purge
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [util] random A B
Generate a random integer in the range [A,B). This is just a command
interface to Python's random.randrange() function.
Arguments:
A start of the range interval (inclusive)
B end of the random range (exclusive, so must be > A)
Options:
-h, --help show this help message and exit
Usage: cylc [db] refresh [OPTIONS] [REGEX]
Check a suite database for invalid registrations (no suite definition
directory or suite.rc file) and refresh suite titles in case they have
changed since the suite was registered. Explicit wildcards must be
used in the match pattern (e.g. 'f' will not match 'foo.bar' unless
you use 'f.⋆').
Arguments:
[REGEX] Suite name match pattern
Options:
-h, --help show this help message and exit
-u, --unregister Automatically unregister invalid registrations.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location, defaults
to $HOME/.cylc/REGDB.
Usage: cylc [db] register [OPTIONS] REG PATH
Register the suite definition located in PATH as REG.
Suite names are hierarchical, delimited by '.' (foo.bar.baz); they
may contain letters, digits, underscore, and hyphens. Colons are not
allowed because directory paths incorporating the suite name are
sometimes needed in PATH variables.
EXAMPLES:
For suite definition directories /home/bob/(one,two,three,four):
% cylc db reg bob /home/bob/one
% cylc db reg foo.bag /home/bob/two
% cylc db reg foo.bar.baz /home/bob/three
% cylc db reg foo.bar.waz /home/bob/four
% cylc db pr '^foo' # print in flat form
bob | "Test Suite One" | /home/bob/one
foo.bag | "Test Suite Two" | /home/bob/two
foo.bar.baz | "Test Suite Four" | /home/bob/three
foo.bar.waz | "Test Suite Three" | /home/bob/four
% cylc db pr -t '^foo' # print in tree form
bob "Test Suite One" | /home/bob/one
foo
|-bag "Test Suite Two" | /home/bob/two
‘-bar
|-baz "Test Suite Three" | /home/bob/three
‘-waz "Test Suite Four" | /home/bob/four
Arguments:
REG Suite name
PATH Suite definition directory
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [control] release|unhold [OPTIONS] REG [MATCH] [POINT]
Release one or more held tasks, or a whole suite. Held tasks do not
submit even if they are ready to run.
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
See also 'cylc [control] hold'.
Arguments:
REG Suite name
[MATCH] Task or family name matching regular expression
[POINT] Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [control] reload [OPTIONS] REG
Tell a suite to reload its definition at run time. All settings
including task definitions, with the exception of suite log
configuration, can be changed on reload. Note that defined tasks can be
be added to or removed from a running suite with the 'cylc insert' and
'cylc remove' commands, without reloading. This command also allows
addition and removal of actual task definitions, and therefore insertion
of tasks that were not defined at all when the suite started (you will
still need to manually insert a particular instance of a newly defined
task). Live task proxies that are orphaned by a reload (i.e. their task
definitions have been removed) will be removed from the task pool if
they have not started running yet. Changes to task definitions take
effect immediately, unless a task is already running at reload time.
If the suite was started with Jinja2 template variables set on the
command line (cylc run --set FOO=bar REG) the same template settings
apply to the reload (only changes to the suite.rc file itself are
reloaded).
If the modified suite definition does not parse, failure to reload will
be reported but no harm will be done to the running suite.
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [control] remove [OPTIONS] REG [MATCH] [POINT]
Remove one or more tasks, or all tasks with a common cycle point from a
running suite.
Tasks will spawn successors first if they have not done so already.
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
Arguments:
REG Suite name
[MATCH] Task or family name matching regular expression
[POINT] Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
--no-spawn Do not spawn successors before removal.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [db] reregister|rename [OPTIONS] REG1 REG2
Change the name of a suite (or group of suites) from REG1 to REG2.
Example:
cylc db rereg foo.bar.baz test.baz
Arguments:
REG1 original name
REG2 new name
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [control] reset [OPTIONS] REG MATCH [POINT]
Reset one or more tasks in a running suite to one of the following states:
⋆ 'waiting' .... prerequisites not satisfied
⋆ 'ready' ...... prerequisites satisfied
⋆ 'succeeded' .. outputs completed
⋆ 'failed' ..... failed
Additionally you can choose:
⋆ 'spawn' ...... force tasks to spawn if they haven't done so already
Tasks set to 'ready' will trigger immediately (see also "cylc trigger").
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
Arguments:
REG Suite name
MATCH Task or family name matching regular expression
[POINT] Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
-s STATE, --state=STATE
Reset task state to STATE, must be one of waiting
ready succeeded failed spawn
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands.
The default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that
it is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [control] restart [OPTIONS] REG [FILE]
Start a suite run from a previous state. To start from scratch (cold or warm
start) see the 'cylc run' command.
The scheduler runs in daemon mode unless you specify n/--no-detach or --debug.
The most recent previous suite state is loaded by default, but earlier state
files in the suite state directory can be specified on the command line.
Tasks recorded as 'submitted' or 'running' will be polled at start-up to
determine what happened to them while the suite was down.
Arguments:
REG Suite name
[FILE] Optional state dump file, assumed to reside in the
suite state dump directory unless an absolute path
is given. Defaults to the most recent suite state.
Options:
-h, --help show this help message and exit
--non-daemon (deprecated: use --no-detach)
-n, --no-detach Do not daemonize the suite
--profile Output profiling (performance) information
--ignore-final-cycle-point
Ignore the final cycle point in the state dump. If one
isspecified in the suite definition it will be used,
however.
--ignore-initial-cycle-point
Ignore the initial cycle point in the state dump. If
one is specified in the suite definition it will be
used, however.
--until=CYCLE_POINT Shut down after all tasks have PASSED this cycle
point.
--hold Hold (don't run tasks) immediately on starting.
--hold-after=CYCLE_POINT
Hold (don't run tasks) AFTER this cycle point.
-m STRING, --mode=STRING
Run mode: live, simulation, or dummy; default is live.
--reference-log Generate a reference log for use in reference tests.
--reference-test Do a test run against a previously generated reference
log.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [control] run|start [OPTIONS] REG [START_POINT]
Start a suite run from scratch, wiping out any previous suite state. To
restart from a previous state see 'cylc restart --help'.
The scheduler runs in daemon mode unless you specify --no-detach or --debug.
Any dependence on cycle points earlier than the start cycle point is ignored.
A "cold start" (the default) starts from the suite's initial cycle point
(specified in the suite.rc or on the command line), and loads any special
one-off cold-start tasks (see below).
A "warm start" (-w/--warm) starts from a given cycle point that is later than
the initial cycle point (specified in the suite.rc), and loads any cold-start
tasks in the 'succeeded' state just to satisfy initial dependence on them.
Aside from the starting cycle point there is no difference between cold and
warm start unless you use special cold-start tasks. See "Suite Start-up" and
"Cold-Start Tasks" in the User Guide for more.
Arguments:
REG Suite name
[START_POINT] Initial cycle point or 'now'; overrides the
suite definition.
Options:
-h, --help show this help message and exit
--non-daemon (deprecated: use --no-detach)
-n, --no-detach Do not daemonize the suite
--profile Output profiling (performance) information
-w, --warm Warm start the suite. The default is to cold start.
--ict Set $CYLC_SUITE_INITIAL_CYCLE_POINT to the initial
cycle point even in a warm start (as for cold starts).
--until=CYCLE_POINT Shut down after all tasks have PASSED this cycle
point.
--hold Hold (don't run tasks) immediately on starting.
--hold-after=CYCLE_POINT
Hold (don't run tasks) AFTER this cycle point.
-m STRING, --mode=STRING
Run mode: live, simulation, or dummy; default is live.
--reference-log Generate a reference log for use in reference tests.
--reference-test Do a test run against a previously generated reference
log.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [discovery] scan [OPTIONS] [HOSTS ...]
Detect running suites by port scanning. Use --verbose to see (with "connection
denied") suites owned by others too.
WARNING: a suite suspended with Ctrl-Z will cause port scans to hang for all
users until timing out (see --pyro-timeout, default 60 seconds).
Arguments:
[HOSTS ...] hosts to scan instead of the default ones
Options:
-h, --help show this help message and exit
--print-ports Print the port range from the site config file
($CYLC_DIR/conf/global.rc).
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to running suites.
The default is 60 seconds.
Usage: cylc [util] scp-transfer [OPTIONS]
An scp wrapper for transferring a list of files and/or directories
at once. The source and target scp URLs can be local or remote (scp
can transfer files between two remote hosts). Passwordless ssh must
be configured appropriately.
ENVIRONMENT VARIABLE INPUTS:
$SRCE - list of sources (files or directories) as scp URLs.
$DEST - parallel list of targets as scp URLs.
The source and destination lists should be space-separated.
We let scp determine the validity of source and target URLs.
Target directories are created pre-copy if they don't exist.
Options:
-v - verbose: print scp stdout.
--help - print this usage message.
Usage: cylc [prep] search|grep [OPTIONS] SUITE PATTERN [PATTERN2...]
Search for pattern matches in suite definitions and any files in the
suite bin directory. Matches are reported by line number and suite
section. An unquoted list of PATTERNs will be converted to an OR'd
pattern. Note that the order of command line arguments conforms to
normal cylc command usage (suite name first) not that of the grep
command.
Note that this command performs a text search on the suite definition,
it does not search the data structure that results from parsing the
suite definition - so it will not report implicit default settings.
For case insenstive matching use '(?i)PATTERN'.
Arguments:
SUITE Suite name or path
PATTERN Python-style regular expression
[PATTERN2...] Additional search patterns
Options:
-h, --help show this help message and exit
-x Do not search in the suite bin directory
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc [control] set-runahead [OPTIONS] REG [HOURS]
Change the suite runahead limit in a running suite. This is the number of
hours that the fastest task is allowed to get ahead of the slowest. If a
task spawns beyond that limit it will be held back from running until the
slowest tasks catch up enough. WARNING: if you omit HOURS no runahead
limit will be set - DO NOT DO THIS for for any cycling suite that has
no near stop cycle set and is not constrained by clock-triggered
tasks.
Arguments:
REG Suite name
[HOURS] Runahead limit (default: no limit)
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [control] set-verbosity [OPTIONS] REG LEVEL
Change the logging priority level of a running suite. Only messages at
or above the chosen priority level will be logged; for example, if you
choose WARNING, only warnings and critical messages will be logged.
Arguments:
REG Suite name
LEVEL INFO, WARNING, NORMAL, CRITICAL, ERROR, DEBUG
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [info] show [OPTIONS] REG [NAME[.CYCLE_POINT]]
Interrogate a running suite for its title and task list, task
descriptions, current state of task prerequisites and outputs and, for
clock-triggered tasks, whether or not the trigger time is up yet.
Arguments:
REG Suite name
[NAME[.CYCLE_POINT]] Task name or ID
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
Usage: cylc [task] started [OPTIONS]
This command is part of the cylc task messaging interface, used by
running tasks to communicate progress to their parent suite.
The started command reports commencement of task execution. It is
called automatically by task job scripts.
Suite and task identity are determined from the task execution
environment supplied by the suite (or by the single task 'submit'
command, in which case the message is just printed to stdout).
See also:
cylc [task] message
cylc [task] succeeded
cylc [task] failed
Options:
-h, --help show this help message and exit
-v, --verbose Verbose output mode.
Usage: cylc [control] stop|shutdown [OPTIONS] REG [STOP]
Tell a running suite daemon to shut down. In order to prevent failures going
unnoticed, suites only shut down automatically at a final cycle point if no
failed tasks are present. There are several shutdown methods (note that an
'active' task is either 'submitted' or 'running'):
1. (default) stop after current active tasks finish
2. (--now) stop immediately, orphaning current active tasks
3. (--kill) stop after killing current active tasks
4. (with STOP as a cycle point) stop after cycle point STOP
5. (with STOP as a task ID) stop after task ID STOP has succeeded
6. (--wall-clock=T) stop after time T (an ISO 8601 date-time format e.g.
CCYYMMDDThh:mm, CCYY-MM-DDThh, etc).
Tasks that become 'ready' after the shutdown is ordered will remain that way
until the shutdown; they will submit if the suite is restarted. Remaining task
event handlers and job poll and kill commands, however, will be executed prior
to shutdown, unless --now is used.
This command exits immediately unless --max-polls is greater than zero, in
which case it polls to wait for suite shutdown.
Arguments:
REG Suite name
[STOP] a/ task POINT (cycle point), or
b/ ISO 8601 date-time (clock time), or
c/ TASK (task ID).
Options:
-h, --help show this help message and exit
-k, --kill Shut down after killing currently active tasks.
-n, --now Shut down immediately, orphaning currently active
tasks.
-w STOP, --wall-clock=STOP
Shut down after time STOP (ISO 8601 formatted)
--max-polls=INT Maximum number of polls (default 0).
--interval=SECS Polling interval in seconds (default 60).
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands.
The default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that
it is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
Usage: cylc [task] submit|single [OPTIONS] REG TASK
Submit a single task to run just as it would be submitted by its suite. Task
messaging commands will print to stdout but will not attempt to communicate
with the suite (which does not need to be running).
For tasks present in the suite graph the given cycle point is adjusted up to
the next valid cycle point for the task. For tasks defined under runtime but
not present in the graph, the given cycle point is assumed to be valid.
WARNING: do not 'cylc submit' a task that is running in its suite at the
same time - both instances will attempt to write to the same job logs.
Arguments:
REG Suite name
TASK Target task (NAME.CYCLE_POINT)
Options:
-h, --help show this help message and exit
-d, --dry-run Generate the cylc task execution file for the task and
report how it would be submitted to run.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [task] succeeded [OPTIONS]
This command is part of the cylc task messaging interface, used by
running tasks to communicate progress to their parent suite.
The succeeded command reports successful completion of task execution.
It is called automatically by task job scripts, except in the case of
"detaching tasks" which must do their own completion messaging.
Suite and task identity are determined from the task execution
environment supplied by the suite (or by the single task 'submit'
command, in which case the message is just printed to stdout).
See also:
cylc [task] message
cylc [task] started
cylc [task] failed
Options:
-h, --help show this help message and exit
-v, --verbose Verbose output mode.
Usage: cylc suite-state REG [OPTIONS]
Print task states retrieved from a suite database; or (with --task,
--point, and --status) poll until a given task reaches a given state.
Polling is configurable with --interval and --max-polls; for a one-off
check use --max-polls=1. The suite database does not need to exist at
the time polling commences but allocated polls are consumed waiting for
it (consider max-polls⋆interval as an overall timeout).
Note for non-cycling tasks --point=1 must be provided.
For your own suites the database location is determined by your
site/user config. For other suites, e.g. those owned by others, or
mirrored suite databases, use --run-dir=DIR to specify the location.
Example usage:
cylc suite-state REG --task=TASK --cycle=CYCLE --status=STATUS
returns 0 if TASK.CYCLE reaches STATUS before the maximum number of
polls, otherwise returns 1.
Arguments:
REG Suite name
Options:
-h, --help show this help message and exit
-t TASK, --task=TASK Specify a task to check the state of.
-c CYCLE, --cycle=CYCLE
Deprecated. Specify the cycle to check task states
for.
-p CYCLE, --point=CYCLE
Specify the cycle point to check task states for.
-d DIR, --run-dir=DIR
The top level cylc run directory if non-standard. The
database should be DIR/REG/cylc-suite.db. Use to
interrogate suites owned by others, etc.; see note
above.
-S STATUS, --status=STATUS
Specify a particular status or triggering condition to
check for. Valid triggering conditions to check for
include: 'fail', 'finish', 'start', 'submit' and
'succeed'. Valid states to check for include:
'failed', 'held', 'queued', 'ready', 'retrying',
'running', 'submit-failed', 'submit-retrying',
'submitted', 'succeeded' and 'waiting'.
--max-polls=INT Maximum number of polls (default 10).
--interval=SECS Polling interval in seconds (default 60).
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
USAGE: cylc test-battery [OPTIONS] [FILES or DIRECTORIES] -- [prove OPTIONS]
Run automated cylc and parsec tests under [FILES or DIRECTORIES].
Test locations default to the following directory tree:
/home/oliverh/cylc/cylc.git/tests/
Some tests (e.g. those specific to particular batch schedulers) can be
configured in your site/user config file. A few others still submit jobs
to a user@host account taken from the environment:
$CYLC_TEST_TASK_HOST # default localhost
$CYLC_TEST_TASK_OWNER # defaut $USER
Requirements:
⋆ Passwordless ssh must be configured to task host accounts.
⋆ Some test suites submit jobs to 'at' so atd must be running.
Suite run directories are cleaned up on the suite host (but not on
remote task hosts) for passing tests - otherwise they are left alone.
By default the command uses the number of processors in your system to
determine the number of tests to run in parallel. This may be anti-social
on systems shared with many other users. On slow or loaded
machines this may result in some test failures due to timeouts intended
to catch problems that can prevent a suite from shutting down normally.
In such cases, you can change the amount of concurrency by setting either:
⋆ the "-j N" option, or
⋆ the "process pool size = N" option in the site/user global
configuration.
To output stderr from failed tests to the terminal, export
CYLC_TEST_DEBUG=true before running this command. To perform comparison
tests using "xxdiff -D", export CYLC_TEST_DEBUG_CMP=true.
For more information see "Reference Tests" in the User Guide.
Options:
-h, --help Print this help message and exit.
Options after '--' are passed through to "prove".
E.g. to run N tests at once, in parallel:
cylc test-battery -- -j N
Usage: cylc [control] trigger [OPTIONS] REG MATCH POINT
Manually trigger tasks. Triggering an unqueued task sets it "ready to
run" and queues it for job submission (cylc internal queues). If the
queue is not limited the task will submit immediately, otherwise it will
submit when released by its queue. Triggering a queued task overrides the
queue limiting mechanism and causes the task to submit immediately (be
aware that this results in a greater number of active tasks than the
queue limit specifies).
For matching multiple tasks or families at once note that MATCH is
interpreted as a full regular expression, not a simple shell glob.
Arguments:
REG Suite name
MATCH Task or family name matching regular expression
POINT Task cycle point (e.g. date-time or integer)
Options:
-h, --help show this help message and exit
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation on
the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
--port=INT Suite port number on the suite host. NOTE: this is
retrieved automatically if passwordless ssh is
configured to the suite host.
--use-ssh Use ssh to re-invoke the command on the suite host.
--no-login Do not use a login shell to run remote ssh commands. The
default is to use a login shell.
--pyro-timeout=SEC Set a timeout for network connections to the running
suite. The default is no timeout. For task messaging
connections see site/user config file documentation.
-f, --force Do not ask for confirmation before acting. Note that it
is not necessary to use this option if interactive
command prompts have been disabled in the site/user
config files.
-m, --family Match members of named families rather than tasks.
Usage: cylc [db] unregister [OPTIONS] REGEX
Remove one or more suites from your suite database. The REGEX pattern
must match whole suite names to avoid accidental de-registration of
partial matches (e.g. 'bar.baz' will not match 'foo.bar.baz').
Associated suite definition directories will not be deleted unless the
'-d,--delete' option is used.
Arguments:
REGEX Regular expression to match suite names.
Options:
-h, --help show this help message and exit
-d, --delete Delete the suite definition directory too (!DANGEROUS!).
-f, --force Don't ask for confirmation before deleting suite definitions.
--user=USER Other user account name. This results in command reinvocation
on the remote account.
--host=HOST Other host name. This results in command reinvocation on the
remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception tracebacks.
--db=PATH Alternative suite registration database location, defaults to
$HOME/.cylc/REGDB.
Usage: cylc upgrade-db
Upgrade a pre-cylc-5.4 suite name database to the new cylc-5.4+ format.
This will create a new-format DB if necessary, or if one already exists
it will transfer old registrations to the new DB so long as the suite
names do not conflict. It is safe to run this utility multiple times.
Prior to cylc-5.4 the suite name registration DB was a Python pickle
file stored at $HOME/.cylc/DB. Since cylc-5.4 it is a directory
$HOME/.cylc/REGDB/ containing one file per registered suite. The
filenames are the suite names, and the file contains key=value pairs:
shell$ cat $HOME/.cylc/REGDB/my.suite
title=my suite title
path=/path/to/my/suite/
Options:
-h, --help show this help message and exit
--from=PATH Path to pre-cylc-5.4 db; default:/home/oliverh/.cylc/DB
--to=PATH Path to new cylc-5.4+ db; default:/home/oliverh/.cylc/REGDB
Usage: cylc [prep] validate [OPTIONS] SUITE
Validate a suite definition against the official specification
files held in $CYLC_DIR/conf/suiterc/.
If the suite definition uses include-files reported line numbers
will correspond to the inlined version seen by the parser; use
'cylc view -i,--inline SUITE' for comparison.
Arguments:
SUITE Suite name or path
Options:
-h, --help show this help message and exit
--ict=ICT Set an initial cycle time to validate against. This
may be required if the suite does not supply one.
--strict Fail any use of unsafe or experimental features.
Currently this just means naked dummy tasks (tasks
with no corresponding runtime section) as these may
result from unintentional typographic errors in task
names.
--no-write Don't write out the processed suite definition.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
Usage: cylc [prep] view [OPTIONS] SUITE
View a read-only temporary copy of suite NAME's suite.rc file, in your
editor, after optional include-file inlining and Jinja2 preprocessing.
The edit process is spawned in the foreground as follows:
% <editor> suite.rc
Where <editor> is defined in the cylc site and user config files
($CYLC_DIR/conf/global.rc and $HOME/.cylc/global.rc).
For remote host or owner, the suite will be printed to stdout unless
the '-g,--gui' flag is used to spawn a remote GUI edit session.
See also 'cylc [prep] edit'.
Arguments:
SUITE Suite name or path
Options:
-h, --help show this help message and exit
-i, --inline Inline include-files.
-j, --jinja2 View after Jinja2 template processing (implies
'-i/--inline' as well).
-p, --process View after all processing (Jinja2, inlining, line-
continuation joining).
-m, --mark (With '-i') Mark inclusions in the left margin.
-l, --label (With '-i') Label file inclusions with the file name.
Line numbers will not correspond to those reported by
the parser.
--single (With '-i') Inline only the first instances of any
multiply-included files. Line numbers will not
correspond to those reported by the parser.
-c, --cat Concatenate continuation lines (line numbers will not
correspond to those reported by the parser).
-g, --gui Force use of the configured GUI editor.
--stdout Print the suite definition to stdout.
--mark-for-edit (With '-i') View file inclusion markers as for 'cylc
edit --inline'.
--user=USER Other user account name. This results in command
reinvocation on the remote account.
--host=HOST Other host name. This results in command reinvocation
on the remote account.
-v, --verbose Verbose output mode.
--debug Run suites in non-daemon mode, and show exception
tracebacks.
--db=PATH Alternative suite registration database location,
defaults to $HOME/.cylc/REGDB.
-s NAME=VALUE, --set=NAME=VALUE
Set the value of a Jinja2 template variable in the
suite definition. This option can be used multiple
times on the command line. WARNING: these settings do
not persist across suite restarts; they need to be set
again on the "cylc restart" command line.
--set-file=FILE Set the value of Jinja2 template variables in the
suite definition from a file containing NAME=VALUE
pairs (one per line). WARNING: these settings do not
persist across suite restarts; they need to be set
again on the "cylc restart" command line.
The graph view in the gcylc GUI shows the structure of the suite as it evolves. It can work well even for large suites, but be aware that the graphviz layout engine has to do a new global layout every time a task proxy appears in or disappears from the task pool. The following may help mitigate any jumping layout problems:
#C: THIS FILE IS PART OF THE CYLC SUITE ENGINE.
#C: Copyright (C) 2008-2014 Hilary Oliver, NIWA
#C:
#C: This program is free software: you can redistribute it and/or modify
#C: it under the terms of the GNU General Public License as published by
#C: the Free Software Foundation, either version 3 of the License, or
#C: (at your option) any later version.
#C:
#C: This program is distributed in the hope that it will be useful,
#C: but WITHOUT ANY WARRANTY; without even the implied warranty of
#C: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
#C: GNU General Public License for more details.
#C:
#C: You should have received a copy of the GNU General Public License
#C: along with this program. If not, see <http://www.gnu.org/licenses/>.
This is the Cylc Suite Engine, version: cylc -v
Access to cylc:
% export PATH=/path/to/cylc/bin:$PATH
% cylc help
% gcylc &
Documentation:
⋆ Installation: /path/to/cylc/INSTALL
⋆ User Guide: /path/to/cylc/doc/index.html
⋆ Project Home Page: http://cylc.github.com/cylc
Code Contributors (git shortlog -s -n):
⋆ Hilary Oliver
⋆ Matt Shin
⋆ Dave Matthews
⋆ Ben Fitzpatrick
⋆ Andrew Clark
⋆ Luis Kornblueh
⋆ Scott Wales
⋆ Kevin Pulo
⋆ Annette Osprey
⋆ Tim Whitcomb
⋆ Alex Reinecke
_______________________________________________________________________
INSTALLING CYLC FROM A SOURCE TARBALL
Cylc is typically installed into a designated "cylc admin" user account
(/home/admin/ below, for illustrative purposes).
% tar xzf cylc-x.y.z.tar.gz
% cd cylc-x.y.z
% export PATH=$PWD/bin:$PATH
% make # (see below to understand what 'make' does here)
_______________________________________________________________________
ARRANGING ACCESS TO CYLC ONCE INSTALLED
To use cylc you just need the cylc bin directory in your $PATH (on task
host accounts as well as on the suite host). Instead of arranging access
to a specific cylc bin directory, however, you should consider
installing successive versions in parallel like this:
/home/admin/cylc/cylc-5.2.0/
/home/admin/cylc/cylc-5.4.0/
# etc.
/home/admin/cylc/cylc-5.4.5/
/home/admin/cylc/cylc -> cylc-5.4.5 # symlink to default version
Now install admin/cylc-wrapper to a convenient location as "cylc", e.g.:
% cp admin/cylc-wrapper /home/admin/bin/cylc
# OR:
% cp admin/cylc-wrapper /usr/local/bin/cylc
Modify the wrapper for your environment as per instructions in the file
and ensure that it is in $PATH for users. The wrapper selects from the
installed versions via an environment variable $CYLC_VERSION that users
can set if they don't want the default version. Running suites also use
this to ensure that their tasks access the right version of cylc.
_______________________________________________________________________
CYLC DEVELOPMENT - CLONING THE GIT REPOSITORY
1) To clone the cylc repository:
% git clone git://github.com/cylc/cylc.git
% cd cylc
% make # build ordereddict and User Guide (as above)
% git pull origin master # update official changes
% make # remake documentation after changes
2) To participate in cylc development: fork cylc on github, clone your
own fork locally, commit changes in feature branches, push them to
your fork and issue a pull request to the cylc development team.
_______________________________________________________________________
WHAT HAPPENS WHEN YOU TYPE 'make'
1) a VERSION file is created containing the cylc version string, e.g.
5.1.0. This is taken from the name of the parent directory - DO NOT
CHANGE THE NAME OF THE UNPACKED SOURCE TREE before running 'make'.
2) the Cylc User Guide is generated from LaTeX source files in doc/:
⋆ if you have pdflatex installed, a PDF version is generated
⋆ if you have tex4ht and ImageMagick convert installed, two HTML
versions (single- and multi-page) are generated
⋆ a doc/index.html is created with links to the generated docs.
3) A Python module "orrdereddict" will be built from source in
ext/ordereddict-0.4.5. To make it available to cylc you must install
the built module into your $PYTHONPATH. This is not essential as a
Python implementation of the Ordered Dictionary structure will be used
by cylc if necessary.
Pyro (Python Remote Objects) is a widely used open source objected oriented Remote Procedure Call technology developed by Irmen de Jong.
Earlier versions of cylc used the Pyro Nameserver to marshal communication between client programs (tasks, commands, viewers, etc.) and their target suites. This worked well, but in principle it provided a route for one suite or user on the network to bring down all running suites by killing the nameserver. Consequently cylc now uses Pyro simply as a lightweight object oriented wrapper for direct network socket communication between client programs and their target suites - all suites are thus entirely isolated from one another.
Cylc 6 introduced new date-time-related syntax for the suite.rc file. In some places, this is quite radically different from the earlier syntax.
Timeouts and delays such as [cylc][[event hooks]]timeout or [runtime][[my_task]]retry delays were written in a purely numeric form before cylc 6, in seconds, minutes (most common), or hours, depending on the setting.
They are now written in an ISO 8601 duration form, which has the benefit that the units are user-selectable (use 1 day instead of 1440 minutes) and explicit.
Nearly all timeouts and delays in cylc were in minutes, except for:
[runtime][[my_task]][[[suite state polling]]]interval
[runtime][[my_task]][[[simulation mode]]]run time range
which were in seconds, and
[scheduling]runahead limit
which was in hours (this is a special case discussed below in J.2).
See Table 1.
|
See A.3.4.
The [scheduling]runahead limit setting was written as a number of hours in pre-cylc-6 suites. This is now in ISO 8601 format for date-time cycling suites, so [scheduling]runahead limit=36 would be written [scheduling]runahead limit=PT36H.
There is a new preferred alternative to runahead limit, [scheduling]max active cycle points. This allows the user to configure how many cycle points can run at once (default 3). See A.3.5.
See A.3.2.
The following suite.rc settings have changed name (Table 2):
|
This change is to reflect the fact that cycling in cylc 6+ can now be over e.g. integers instead of being purely based on date-time.
Date-times written in initial cycle time and final cycle time were in a cylc-specific 10-digit (or less) CCYYMMDDhh format, such as 2014021400 for 00:00 on the 14th of February 2014.
Date-times are now required to be ISO 8601 compatible. This can be achieved easily enough by inserting a T between the day and the hour digits.
|
Start-up tasks have been removed from cylc 6, and use of cold-start tasks is no longer recommended. Instead, you should use the initial/repeat-once notation as detailed in 7.23.2 and 9.3.3.2.3.
Repeating asynchronous tasks have also been removed because non date-time workflows can now be handled more easily with integer cycling. See for instance the satellite data processing example documented in 9.3.3.2.4.
For repeating tasks with hour-based cycling the syntax has only minor changes:
Pre-cylc-6:
Hour-based cycling section names are easy enough to convert, as seen in Table 4.
|
The graph text in hour-based cycling is also easy to convert, as seen in Table 5.
|
Prior to cylc-6 inter-cycle triggers implicitly created task instances at the offset cycle points. For example, this pre cylc-6 suite automatically creates instances of task foo at the offset hours 3,9,15,21 each day, for task bar to trigger off at 0,6,12,18:
You can run this suite to see how it works (cylc-6+ is backward-compatible: it generates the old-style behaviour when it detects an old-style suite definition). Here’s the direct translation of this suite to cylc-6+ format:
If you run this suite, bar.20140808T00 will execute at start-up because dependence on tasks prior to the initial cycle point is ignored, but the suite will then stall with bar.20140808T06 waiting on foo.20140808T03 (right-click on the task in the GUI to see its prerequisites), which does not exist because the offset foo instances have not been created. Note that if you graph the suite the offset foo instances do appear. That’s because the graph expresses dependence, and bar really does depend on these instances of foo. The problem is just that the tasks don’t get created at run time, which is why the suite stalls. Here’s the correct way to get the desired behaviour in cylc-6+:
Implicit creation of task instances by offset triggers has been disabled because it is error prone: if a trigger offset is wrong it should cause a triggering failure rather than create task instances at incorrect cycle points.
Copyright Ⓒ 2007 Free Software Foundation, Inc. http://fsf.org/
Everyone is permitted to copy and distribute verbatim copies of this
license document, but changing it is not allowed.
Preamble
The GNU General Public License is a free, copyleft license for software and other kinds of works.
The licenses for most software and other practical works are designed to take away your freedom to share and change the works. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change all versions of a program–to make sure it remains free software for all its users. We, the Free Software Foundation, use the GNU General Public License for most of our software; it applies also to any other work released this way by its authors. You can apply it to your programs, too.
When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for them if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs, and that you know you can do these things.
To protect your rights, we need to prevent others from denying you these rights or asking you to surrender the rights. Therefore, you have certain responsibilities if you distribute copies of the software, or if you modify it: responsibilities to respect the freedom of others.
For example, if you distribute copies of such a program, whether gratis or for a fee, you must pass on to the recipients the same freedoms that you received. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.
Developers that use the GNU GPL protect your rights with two steps: (1) assert copyright on the software, and (2) offer you this License giving you legal permission to copy, distribute and/or modify it.
For the developers’ and authors’ protection, the GPL clearly explains that there is no warranty for this free software. For both users’ and authors’ sake, the GPL requires that modified versions be marked as changed, so that their problems will not be attributed erroneously to authors of previous versions.
Some devices are designed to deny users access to install or run modified versions of the software inside them, although the manufacturer can do so. This is fundamentally incompatible with the aim of protecting users’ freedom to change the software. The systematic pattern of such abuse occurs in the area of products for individuals to use, which is precisely where it is most unacceptable. Therefore, we have designed this version of the GPL to prohibit the practice for those products. If such problems arise substantially in other domains, we stand ready to extend this provision to those domains in future versions of the GPL, as needed to protect the freedom of users.
Finally, every program is threatened constantly by software patents. States should not allow patents to restrict development and use of software on general-purpose computers, but in those that do, we wish to avoid the special danger that patents applied to a free program could make it effectively proprietary. To prevent this, the GPL assures that patents cannot be used to render the program non-free.
The precise terms and conditions for copying, distribution and modification follow.
Terms and Conditions
“This License” refers to version 3 of the GNU General Public License.
“Copyright” also means copyright-like laws that apply to other kinds of works, such as semiconductor masks.
“The Program” refers to any copyrightable work licensed under this License. Each licensee is addressed as “you”. “Licensees” and “recipients” may be individuals or organizations.
To “modify” a work means to copy from or adapt all or part of the work in a fashion requiring copyright permission, other than the making of an exact copy. The resulting work is called a “modified version” of the earlier work or a work “based on” the earlier work.
A “covered work” means either the unmodified Program or a work based on the Program.
To “propagate” a work means to do anything with it that, without permission, would make you directly or secondarily liable for infringement under applicable copyright law, except executing it on a computer or modifying a private copy. Propagation includes copying, distribution (with or without modification), making available to the public, and in some countries other activities as well.
To “convey” a work means any kind of propagation that enables other parties to make or receive copies. Mere interaction with a user through a computer network, with no transfer of a copy, is not conveying.
An interactive user interface displays “Appropriate Legal Notices” to the extent that it includes a convenient and prominently visible feature that (1) displays an appropriate copyright notice, and (2) tells the user that there is no warranty for the work (except to the extent that warranties are provided), that licensees may convey the work under this License, and how to view a copy of this License. If the interface presents a list of user commands or options, such as a menu, a prominent item in the list meets this criterion.
The “source code” for a work means the preferred form of the work for making modifications to it. “Object code” means any non-source form of a work.
A “Standard Interface” means an interface that either is an official standard defined by a recognized standards body, or, in the case of interfaces specified for a particular programming language, one that is widely used among developers working in that language.
The “System Libraries” of an executable work include anything, other than the work as a whole, that (a) is included in the normal form of packaging a Major Component, but which is not part of that Major Component, and (b) serves only to enable use of the work with that Major Component, or to implement a Standard Interface for which an implementation is available to the public in source code form. A “Major Component”, in this context, means a major essential component (kernel, window system, and so on) of the specific operating system (if any) on which the executable work runs, or a compiler used to produce the work, or an object code interpreter used to run it.
The “Corresponding Source” for a work in object code form means all the source code needed to generate, install, and (for an executable work) run the object code and to modify the work, including scripts to control those activities. However, it does not include the work’s System Libraries, or general-purpose tools or generally available free programs which are used unmodified in performing those activities but which are not part of the work. For example, Corresponding Source includes interface definition files associated with source files for the work, and the source code for shared libraries and dynamically linked subprograms that the work is specifically designed to require, such as by intimate data communication or control flow between those subprograms and other parts of the work.
The Corresponding Source need not include anything that users can regenerate automatically from other parts of the Corresponding Source.
The Corresponding Source for a work in source code form is that same work.
All rights granted under this License are granted for the term of copyright on the Program, and are irrevocable provided the stated conditions are met. This License explicitly affirms your unlimited permission to run the unmodified Program. The output from running a covered work is covered by this License only if the output, given its content, constitutes a covered work. This License acknowledges your rights of fair use or other equivalent, as provided by copyright law.
You may make, run and propagate covered works that you do not convey, without conditions so long as your license otherwise remains in force. You may convey covered works to others for the sole purpose of having them make modifications exclusively for you, or provide you with facilities for running those works, provided that you comply with the terms of this License in conveying all material for which you do not control copyright. Those thus making or running the covered works for you must do so exclusively on your behalf, under your direction and control, on terms that prohibit them from making any copies of your copyrighted material outside their relationship with you.
Conveying under any other circumstances is permitted solely under the conditions stated below. Sublicensing is not allowed; section 10 makes it unnecessary.
No covered work shall be deemed part of an effective technological measure under any applicable law fulfilling obligations under article 11 of the WIPO copyright treaty adopted on 20 December 1996, or similar laws prohibiting or restricting circumvention of such measures.
When you convey a covered work, you waive any legal power to forbid circumvention of technological measures to the extent such circumvention is effected by exercising rights under this License with respect to the covered work, and you disclaim any intention to limit operation or modification of the work as a means of enforcing, against the work’s users, your or third parties’ legal rights to forbid circumvention of technological measures.
You may convey verbatim copies of the Program’s source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice; keep intact all notices stating that this License and any non-permissive terms added in accord with section 7 apply to the code; keep intact all notices of the absence of any warranty; and give all recipients a copy of this License along with the Program.
You may charge any price or no price for each copy that you convey, and you may offer support or warranty protection for a fee.
You may convey a work based on the Program, or the modifications to produce it from the Program, in the form of source code under the terms of section 4, provided that you also meet all of these conditions:
A compilation of a covered work with other separate and independent works, which are not by their nature extensions of the covered work, and which are not combined with it such as to form a larger program, in or on a volume of a storage or distribution medium, is called an “aggregate” if the compilation and its resulting copyright are not used to limit the access or legal rights of the compilation’s users beyond what the individual works permit. Inclusion of a covered work in an aggregate does not cause this License to apply to the other parts of the aggregate.
You may convey a covered work in object code form under the terms of sections 4 and 5, provided that you also convey the machine-readable Corresponding Source under the terms of this License, in one of these ways:
A separable portion of the object code, whose source code is excluded from the Corresponding Source as a System Library, need not be included in conveying the object code work.
A “User Product” is either (1) a “consumer product”, which means any tangible personal property which is normally used for personal, family, or household purposes, or (2) anything designed or sold for incorporation into a dwelling. In determining whether a product is a consumer product, doubtful cases shall be resolved in favor of coverage. For a particular product received by a particular user, “normally used” refers to a typical or common use of that class of product, regardless of the status of the particular user or of the way in which the particular user actually uses, or expects or is expected to use, the product. A product is a consumer product regardless of whether the product has substantial commercial, industrial or non-consumer uses, unless such uses represent the only significant mode of use of the product.
“Installation Information” for a User Product means any methods, procedures, authorization keys, or other information required to install and execute modified versions of a covered work in that User Product from a modified version of its Corresponding Source. The information must suffice to ensure that the continued functioning of the modified object code is in no case prevented or interfered with solely because modification has been made.
If you convey an object code work under this section in, or with, or specifically for use in, a User Product, and the conveying occurs as part of a transaction in which the right of possession and use of the User Product is transferred to the recipient in perpetuity or for a fixed term (regardless of how the transaction is characterized), the Corresponding Source conveyed under this section must be accompanied by the Installation Information. But this requirement does not apply if neither you nor any third party retains the ability to install modified object code on the User Product (for example, the work has been installed in ROM).
The requirement to provide Installation Information does not include a requirement to continue to provide support service, warranty, or updates for a work that has been modified or installed by the recipient, or for the User Product in which it has been modified or installed. Access to a network may be denied when the modification itself materially and adversely affects the operation of the network or violates the rules and protocols for communication across the network.
Corresponding Source conveyed, and Installation Information provided, in accord with this section must be in a format that is publicly documented (and with an implementation available to the public in source code form), and must require no special password or key for unpacking, reading or copying.
“Additional permissions” are terms that supplement the terms of this License by making exceptions from one or more of its conditions. Additional permissions that are applicable to the entire Program shall be treated as though they were included in this License, to the extent that they are valid under applicable law. If additional permissions apply only to part of the Program, that part may be used separately under those permissions, but the entire Program remains governed by this License without regard to the additional permissions.
When you convey a copy of a covered work, you may at your option remove any additional permissions from that copy, or from any part of it. (Additional permissions may be written to require their own removal in certain cases when you modify the work.) You may place additional permissions on material, added by you to a covered work, for which you have or can give appropriate copyright permission.
Notwithstanding any other provision of this License, for material you add to a covered work, you may (if authorized by the copyright holders of that material) supplement the terms of this License with terms:
All other non-permissive additional terms are considered “further restrictions” within the meaning of section 10. If the Program as you received it, or any part of it, contains a notice stating that it is governed by this License along with a term that is a further restriction, you may remove that term. If a license document contains a further restriction but permits relicensing or conveying under this License, you may add to a covered work material governed by the terms of that license document, provided that the further restriction does not survive such relicensing or conveying.
If you add terms to a covered work in accord with this section, you must place, in the relevant source files, a statement of the additional terms that apply to those files, or a notice indicating where to find the applicable terms.
Additional terms, permissive or non-permissive, may be stated in the form of a separately written license, or stated as exceptions; the above requirements apply either way.
You may not propagate or modify a covered work except as expressly provided under this License. Any attempt otherwise to propagate or modify it is void, and will automatically terminate your rights under this License (including any patent licenses granted under the third paragraph of section 11).
However, if you cease all violation of this License, then your license from a particular copyright holder is reinstated (a) provisionally, unless and until the copyright holder explicitly and finally terminates your license, and (b) permanently, if the copyright holder fails to notify you of the violation by some reasonable means prior to 60 days after the cessation.
Moreover, your license from a particular copyright holder is reinstated permanently if the copyright holder notifies you of the violation by some reasonable means, this is the first time you have received notice of violation of this License (for any work) from that copyright holder, and you cure the violation prior to 30 days after your receipt of the notice.
Termination of your rights under this section does not terminate the licenses of parties who have received copies or rights from you under this License. If your rights have been terminated and not permanently reinstated, you do not qualify to receive new licenses for the same material under section 10.
You are not required to accept this License in order to receive or run a copy of the Program. Ancillary propagation of a covered work occurring solely as a consequence of using peer-to-peer transmission to receive a copy likewise does not require acceptance. However, nothing other than this License grants you permission to propagate or modify any covered work. These actions infringe copyright if you do not accept this License. Therefore, by modifying or propagating a covered work, you indicate your acceptance of this License to do so.
Each time you convey a covered work, the recipient automatically receives a license from the original licensors, to run, modify and propagate that work, subject to this License. You are not responsible for enforcing compliance by third parties with this License.
An “entity transaction” is a transaction transferring control of an organization, or substantially all assets of one, or subdividing an organization, or merging organizations. If propagation of a covered work results from an entity transaction, each party to that transaction who receives a copy of the work also receives whatever licenses to the work the party’s predecessor in interest had or could give under the previous paragraph, plus a right to possession of the Corresponding Source of the work from the predecessor in interest, if the predecessor has it or can get it with reasonable efforts.
You may not impose any further restrictions on the exercise of the rights granted or affirmed under this License. For example, you may not impose a license fee, royalty, or other charge for exercise of rights granted under this License, and you may not initiate litigation (including a cross-claim or counterclaim in a lawsuit) alleging that any patent claim is infringed by making, using, selling, offering for sale, or importing the Program or any portion of it.
A “contributor” is a copyright holder who authorizes use under this License of the Program or a work on which the Program is based. The work thus licensed is called the contributor’s “contributor version”.
A contributor’s “essential patent claims” are all patent claims owned or controlled by the contributor, whether already acquired or hereafter acquired, that would be infringed by some manner, permitted by this License, of making, using, or selling its contributor version, but do not include claims that would be infringed only as a consequence of further modification of the contributor version. For purposes of this definition, “control” includes the right to grant patent sublicenses in a manner consistent with the requirements of this License.
Each contributor grants you a non-exclusive, worldwide, royalty-free patent license under the contributor’s essential patent claims, to make, use, sell, offer for sale, import and otherwise run, modify and propagate the contents of its contributor version.
In the following three paragraphs, a “patent license” is any express agreement or commitment, however denominated, not to enforce a patent (such as an express permission to practice a patent or covenant not to sue for patent infringement). To “grant” such a patent license to a party means to make such an agreement or commitment not to enforce a patent against the party.
If you convey a covered work, knowingly relying on a patent license, and the Corresponding Source of the work is not available for anyone to copy, free of charge and under the terms of this License, through a publicly available network server or other readily accessible means, then you must either (1) cause the Corresponding Source to be so available, or (2) arrange to deprive yourself of the benefit of the patent license for this particular work, or (3) arrange, in a manner consistent with the requirements of this License, to extend the patent license to downstream recipients. “Knowingly relying” means you have actual knowledge that, but for the patent license, your conveying the covered work in a country, or your recipient’s use of the covered work in a country, would infringe one or more identifiable patents in that country that you have reason to believe are valid.
If, pursuant to or in connection with a single transaction or arrangement, you convey, or propagate by procuring conveyance of, a covered work, and grant a patent license to some of the parties receiving the covered work authorizing them to use, propagate, modify or convey a specific copy of the covered work, then the patent license you grant is automatically extended to all recipients of the covered work and works based on it.
A patent license is “discriminatory” if it does not include within the scope of its coverage, prohibits the exercise of, or is conditioned on the non-exercise of one or more of the rights that are specifically granted under this License. You may not convey a covered work if you are a party to an arrangement with a third party that is in the business of distributing software, under which you make payment to the third party based on the extent of your activity of conveying the work, and under which the third party grants, to any of the parties who would receive the covered work from you, a discriminatory patent license (a) in connection with copies of the covered work conveyed by you (or copies made from those copies), or (b) primarily for and in connection with specific products or compilations that contain the covered work, unless you entered into that arrangement, or that patent license was granted, prior to 28 March 2007.
Nothing in this License shall be construed as excluding or limiting any implied license or other defenses to infringement that may otherwise be available to you under applicable patent law.
If conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot convey a covered work so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not convey it at all. For example, if you agree to terms that obligate you to collect a royalty for further conveying from those to whom you convey the Program, the only way you could satisfy both those terms and this License would be to refrain entirely from conveying the Program.
Notwithstanding any other provision of this License, you have permission to link or combine any covered work with a work licensed under version 3 of the GNU Affero General Public License into a single combined work, and to convey the resulting work. The terms of this License will continue to apply to the part which is the covered work, but the special requirements of the GNU Affero General Public License, section 13, concerning interaction through a network will apply to the combination as such.
The Free Software Foundation may publish revised and/or new versions of the GNU General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the Program specifies that a certain numbered version of the GNU General Public License “or any later version” applies to it, you have the option of following the terms and conditions either of that numbered version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of the GNU General Public License, you may choose any version ever published by the Free Software Foundation.
If the Program specifies that a proxy can decide which future versions of the GNU General Public License can be used, that proxy’s public statement of acceptance of a version permanently authorizes you to choose that version for the Program.
Later license versions may give you additional or different permissions. However, no additional obligations are imposed on any author or copyright holder as a result of your choosing to follow a later version.
THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
If the disclaimer of warranty and limitation of liability provided above cannot be given local legal effect according to their terms, reviewing courts shall apply local law that most closely approximates an absolute waiver of all civil liability in connection with the Program, unless a warranty or assumption of liability accompanies a copy of the Program in return for a fee.
End of Terms and Conditions
How to Apply These Terms to Your New Programs
If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively state the exclusion of warranty; and each file should have at least the “copyright” line and a pointer to where the full notice is found.
<one line to give the program's name and a brief idea of what it does.>
Copyright (C) <textyear> <name of author> This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see <http://www.gnu.org/licenses/>. |
Also add information on how to contact you by electronic and paper mail.
If the program does terminal interaction, make it output a short notice like this when it starts in an interactive mode:
<program> Copyright (C) <year> <name of author>
This program comes with ABSOLUTELY NO WARRANTY; for details type ‘show w'. This is free software, and you are welcome to redistribute it under certain conditions; type ‘show c' for details. |
The hypothetical commands show w and show c should show the appropriate parts of the General Public License. Of course, your program’s commands might be different; for a GUI interface, you would use an “about box”.
You should also get your employer (if you work as a programmer) or school, if any, to sign a “copyright disclaimer” for the program, if necessary. For more information on this, and how to apply and follow the GNU GPL, see http://www.gnu.org/licenses/.
The GNU General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Lesser General Public License instead of this License. But first, please read http://www.gnu.org/philosophy/why-not-lgpl.html.
1Future plans for EcoConnect include additional deterministic regional weather forecasts and a statistical ensemble.
2An OR operator on the right doesn’t make much sense: if “B or C” triggers off A, what exactly should cylc do when A finishes?
3In NWP forecast analysis suites parts of the observation processing and data assimilation subsystem will typically also depend on model background fields generated by the previous forecast.
4Note that $CYLC_SUITE_ENVIRONMENT is a string containing embedded newline characters and it has to be handled accordingly. In the bash shell, for instance, it should be echoed in quotes to avoid concatenation to a single line.
5If you accidentally delete a port file while a suite is running, use cylc scan to determine the port number then use it on the command line (--port) or rewrite the port file manually.